|
somewhat similar to ZBXNEXT-428 (time suffixes for intervals)
one problem is that everything is currently optimised for integers-only intervals
|
|
It's great for some cases to use predefined timerange.
For example: some template,10 hosts with this template. One host have problems. I'm replace (with macros) timeranges for more precision monitoring this host. When problems has gone, I can quiqly change back timeranges.
|
|
The feature will be very useful. Currently hard codding the time intervals is less flexible.
|
|
I realize its not the original request; but for discovery, it would be very useful to be able to put lld macros in interval field.
|
|
Real case:
ISP have network with some layers
Access - need frequently check availibility (icmp ping every 60s) and rarely collect SNMP data (ifTraffic, ifErrors - from 5min up to 24hours for hostname or location)
Distribution - need frequently check availibility (icmp ping every 60s) and more frequently collect more SNMP data
...
Theoretically we can do
Template access = Template ICMP + Template SNMP Device (slow poll interval)
Template distibution = Template ICMP + Template SNMP Device (medium poll interval)
then
define {$SNMP_ifTraffic_PollInterval} = 600 in acccess template
define {$SNMP_ifTraffic_PollInterval} = 300 in distribution template
but we can't.
Therefore we do separate "rock solid" template "template access" and "template distribution". Every template contain copy of all elemets from Template ICMP and Template SNMP Device. And here we can set different update intervals in every template separately. You lose the advantage of the inheritance patterns. Huh? 
|
|
+1
And it will provide possibility to define useful trigger descriptions, like this:
{$ICMP_RETRY_COUNT}=#3
{$ICMP_RETRY_COUNT_THRESH}=3
p.s. two macros because not possible define this function: count(#{$ICMP_RETRY_THRESH},0)}
Expr:
{ZBX-CISCO-ENVMON:icmpping.count({$ICMP_RETRY_COUNT},0)}={$ICMP_RETRY_COUNT_THRESH}
Descr:
Host not responding to last {$ICMP_RETRY_COUNT_THRESH} ICMP polls. Poll interval is {$NORMAL}sec
|
|
Yes, when this will introduced?
|
|
This would also be a very nice place to be abe to use context aware macros.
This way we can define a default macro and override it with context if needed.
|
|
Raymond, that's exactly the plan. The functionality is in our roadmap for Zabbix 3.4.
|
|
Very nice, thanks for the feedback Alexei!
Unfortunately, I don't see this ZBXNEXT on http://zabbix.org/wiki/Docs/roadmap#Zabbix_3.4_roadmap.
Can the list please be updated with the current list of features under development?
|
|
It's also needed for calculation interface utilization, here is cisco formula:
http://www.cisco.com/c/dam/en/us/support/docs/ip/simple-network-management-protocol-snmp/8141-calculate-bandwidth-snmp1.gif
source:
http://www.cisco.com/c/en/us/support/docs/ip/simple-network-management-protocol-snmp/8141-calculate-bandwidth-snmp.html
|
|
Dear romale, Store as set to Delta (speed per second) in item configuration will do the job for you.
|
|
(1) [S] Integer overflow is possible in:
- lld_hosts_remove and lld_groups_remove - called from lld_update_hosts
- lld_remove_lost_items and lld_remove_lost_applications - called from lld_update_items
as lifetime type changed from unsigned short to int with allowed upper limit defined (when checking user input) as 2147483647 (0x7fffffff).
glebs.ivanovskis Current approach relies heavily on frontend-side validation and few more assumptions. I got your point, we need more robust time arithmetic.
glebs.ivanovskis RESOLVED in r65533.
vjaceslavs Thanks! CLOSED
|
|
(2) [S] Flexible interval parsing (flexible_interval_parse) and zbx_check_time_period fails because time_period_parse fails:
There is a missing increment of pointer text in:
When checking time period like "00:00-24:00", first time_parse call is made with text "00:00-24:00", but second time_parse call is made with text "-24:00".
glebs.ivanovskis Thank you! RESOLVED in r65525.
vjaceslavs CLOSED
|
|
(3) [S] When item becomes "not supported" due to invalid update interval it never gets polled again (unless server is restarted). And there are two ways to fix the issue:
- schedule item next check based on "refresh unsupported interval" - this will result in flapping, because item itself may be perfectly fine and once polled it becomes supported but then it cannot be scheduled and so on...
- do a special processing for such "not supported" items - this is not very logical because splits "not supported" items in two types instead of splitting "not supported" state.
So the problem is that "not supported" state does not really suit invalid interval case. Introducing new state might be a tall order.
glebs.ivanovskis Finally RESOLVED in r67780. (There were actually many changes leading to this fix, I can't even name them all. It's probably better to review difference between dbconfig.c in development branch and trunk@<latest merge revision>.)
vjaceslavs Looks like constants are copy-pasted in dbsync_item_preproc_row (lines 1556 - 1563) as there are different flags set by dbsync_check_row_macros, but only one constant (ZBX_DBSYNC_ITEM_COLUMN_DELAY) is used in checks.
REOPENED
glebs.ivanovskis Silly mistake, thank you for keeping an eye on me! RESOLVED in r67810.
vjaceslavs CLOSED
|
|
(4) [A][F][I][PS] There used to be very limited validation of History storage period and Trend storage period in frontend and API. Server should prevent potential overflows in housekeeper by limiting these fields during upgrade.
glebs.ivanovskis However, this shouldn't cause any data loss, just annoying log messages if there are some bogus housekeeping settings (anything up to 24855 days will be fine). Some additional protection on server side is added in r65720. RESOLVED ([PS])
glebs.ivanovskis As discussed, housekeeping settings should be limited to 25 * 365 days. DB upgrade updated in r65729. Still RESOLVED (I mean [I] part.)
sasha RESOLVED [AF] in r65940, r65947
iivs With minor coding style fixes in r66252, r66225
CLOSED
glebs.ivanovskis [I][PS] is still RESOLVED
vjaceslavs CLOSED
|
|
(5) [A][F][I][PS] For the sake on consistency, should we change web scenario step timeout to also support time unit suffixes? The default is 15 seconds. Instead of 15, the user could enter it as 15s. The validation of maximum timeout could be adjusted later. For example maximum could be entered as 1m.
glebs.ivanovskis RESOLVED [I] in r65743 and [PS] in r65764.
sasha RESOLVED [AF] in r66092
iivs With minor coding style fix in r66252
CLOSED
glebs.ivanovskis [IPS] are still RESOLVED
vjaceslavs [I] is ok
[PS] The error message must start from the upper case and end with a dot. REOPENED
glebs.ivanovskis Which messages are you talking about? Web scenario errors like "cURL library is required for Web monitoring support" are not capitalized or ended with a dot as they are part of a longer message like "Step n of m failed: ..."
vjaceslavs I was talking about messages like "update interval "XXX" is invalid", but I get the point. Messages are a part of a longer message, but this message does not end with the dot (Step "bebebe" [1 of 1] failed: update interval "33 cows" is invalid). But I agree, that it is not related to this subissue, so
CLOSED
|
|
(6) [I] As discussed, the filter and subfilter in item list should be cleared on DB upgrade.
Table: profiles
keys:
- web.items.filter_delay
- web.items.filter_history
- web.items.filter_trends
- web.items.subfilter_interval
- web.items.subfilter_history
- web.items.subfilter_trends
glebs.ivanovskis RESOLVED in r65714.
vjaceslavs CLOSED
|
|
(7) [I] Seems like some table columns still have old default values in data.tmpl although the new format is expected.
glebs.ivanovskis There are two main reasons why it hasn't been done yet:
- we need to decide whether we will promote macro usage for update intervals in official templates or limit ourselves with time units;
- I want to postpone items table editing as close to the final merge as possible because of painful conflict resolution with any table edit in trunk (since I've deleted delay_flex column in dev branch).
glebs.ivanovskis RESOLVED in r66079, r66209.
vjaceslavs CLOSED
|
|
(9) [F] Clean install of svn://svn.zabbix.com/branches/dev/ZBXNEXT-1675 r65686, went straight to graphs:
frontends/php/include/graphs.inc.php:164
*/
function get_min_itemclock_by_graphid($graphid) {
$items = DBfetchArray(DBselect(
- 'SELECT DISTINCT i.itemid,i.value_type,i.history,i.trends'.
+ 'SELECT DISTINCT i.itemid,i.value_type,i.history,i.trends,i.hostid'.
' FROM items i,graphs_items gi'.
' WHERE i.itemid=gi.itemid'.
' AND gi.graphid='.zbx_dbstr($graphid)
sasha RESOLVED in r65938
iivs CLOSED
|
|
(10) [FS] Frontend fails to calculate required performance.
As this calculation gets more and more complicated with macros and other stuff, there were rumors that we will introduce a new request, so that frontend could get status information from running server. Please consult with sasha and move this subissue into its own ticket.
glebs.ivanovskis Responding to frontend pings implemented in r66143, getting status information from server's cache is implemented in r66172. [S] RESOLVED
sasha crash occurred while getting user stats
sasha REOPENED
glebs.ivanovskis Crashing issue RESOLVED in r66174.
sasha RESOLVED [F] in r66189
iivs Dashboard widget empty. In screens, the error is more visible:
REOPENED
sasha RESOLVED in r66269
iivs CLOSED
glebs.ivanovskis Improved server-side implementation in r67147. [S] is still RESOLVED
vjaceslavs CLOSED
|
|
(11) [F] Error message can be improved.
iivs The validation is on API side. Same validator used in other places that use the same syntax, so the error message is the same and won't be changed. Discussed with sasha.
WON'T FIX
|
|
(12) [D] API documentation updated:
sasha
Additional changes:
These pages needs to be updated:
REOPENED
iivs RESOLVED
sasha Thanks! CLOSED
|
|
(13) [F] Translation string changes.
Strings added:
- (0 - use action default)
- Audit data storage period "%1$s".
- Auto-logout
- Auto-registration data storage period
- Auto-registration event and alert data storage period "%1$s".
- Data storage period
- Default delay
- History data storage period "%1$s".
- Internal data storage period
- Internal event and alert data storage period "%1$s".
- Invalid IT services data storage period: %1$s.
- Invalid audit data storage period: %1$s.
- Invalid auto-registration data storage period: %1$s.
- Invalid blinking on trigger status change: %1$s.
- Invalid date "%s".
- Invalid displaying of OK triggers: %1$s.
- Invalid event expiry time: %1$s.
- Invalid fifth parameter.
- Invalid first parameter.
- Invalid fourth parameter.
- Invalid history data storage period: %1$s.
- Invalid hostid for macro "%1$s".
- Invalid internal data storage period: %1$s.
- Invalid interval "%1$s".
- Invalid network discovery data storage period: %1$s.
- Invalid refresh of unsupported items: %1$s
- Invalid time period.
- Invalid trends data storage period: %1$s.
- Invalid trigger data storage period: %1$s.
- Invalid user sessions data storage period: %1$s.
- Keep lost resources period
- Network discovery data storage period
- Network discovery event and alert data storage period "%1$s".
- Refresh
- Refresh unsupported items
- Refresh unsupported items "%1$s".
- Service data storage period "%1$s".
- Show all (%1$s)
- Show events not older than
- Show events not older than "%1$s".
- Show unacknowledged (%1$s)
- Trend data storage period "%1$s".
- Trigger data storage period
- Trigger event and alert data storage period "%1$s".
- User session data storage period "%1$s".
- a time unit is expected
- invalid delay
- invalid history storage period
- invalid trend storage period
Strings deleted:
- %1$s day
- %1$s days
- (minimum 60 seconds, 0 - use action default)
- Action "%1$s" has incorrect value for "esc_period" (minimum %2$s seconds).
- Audit data storage period (in days) "%1$s".
- Auto-logout (min 90 seconds)
- Auto-registration data storage period (in days)
- Auto-registration event and alert data storage period (in days) "%1$s".
- Data storage period (in days)
- Default delay (in seconds)
- Delay (in sec)
- Discovery rule "%1$s:%2$s" has incorrect lifetime: "%3$s". (min: %4$d, max: %5$d, user macro allowed)
- Duration (sec)
- Empty time period.
- History (in days)
- History data storage period (in days) "%1$s".
- History storage period (in days)
- Hosts (m/n/t)
- Incorrect action operation step duration.
- Incorrect time period "%1$s" start day must be less or equal to end day.
- Incorrect time period "%1$s" start time must be less than end time.
- Incorrect time period "%1$s".
- Internal data storage period (in days)
- Internal event and alert data storage period (in days) "%1$s".
- Invalid date "%s".
- Invalid fifth parameter.
- Invalid first parameter.
- Invalid flexible interval delay: "%1$s" exceeds maximum delay of "%2$s".
- Invalid fourth parameter.
- Invalid hostid for macro "%1$s".
- Invalid interval "%1$s".
- Invalid interval "%1$s": %2$s.
- Invalid interval "%1$s": invalid hour step "%2$s".
- Invalid interval "%1$s": invalid hours "%2$s".
- Invalid interval "%1$s": invalid minute step "%2$s".
- Invalid interval "%1$s": invalid minutes "%2$s".
- Invalid interval "%1$s": invalid month day "%2$s".
- Invalid interval "%1$s": invalid month day step "%2$s".
- Invalid interval "%1$s": invalid second step "%2$s".
- Invalid interval "%1$s": invalid seconds "%2$s".
- Invalid interval "%1$s": invalid week day "%2$s".
- Invalid interval "%1$s": invalid week day step "%2$s".
- Invalid interval "%1$s": starting hour must be less or equal to ending hour.
- Invalid interval "%1$s": starting minute must be less or equal to ending minute.
- Invalid interval "%1$s": starting month day must be less or equal to ending month day.
- Invalid interval "%1$s": starting second must be less or equal to ending second.
- Invalid interval "%1$s": starting week day must be less or equal to ending week day.
- Items (m/d/n)
- Keep lost resources period (in days)
- Logged in as
- Message timeout (seconds)
- Network discovery data storage period (in days)
- Network discovery event and alert data storage period (in days) "%1$s".
- Refresh (in seconds)
- Refresh unsupported items (in sec)
- Refresh unsupported items (in sec) "%1$s".
- Service data storage period (in days) "%1$s".
- Show all (%1$s day)
- Show all (%1$s days)
- Show events not older than (in days)
- Show events not older than (in days) "%1$s".
- Show unacknowledged (%1$s day)
- Show unacknowledged (%1$s days)
- Trend data storage period (in days) "%1$s".
- Trend storage period (in days)
- Trends (in days)
- Trigger data storage period (in days)
- Trigger event and alert data storage period (in days) "%1$s".
- Triggers (e/d)[p/o]
- Update interval (in sec)
- User session data storage period (in days) "%1$s".
- Users (online)
- is
- minimum 60 seconds
- not running
- running
- sec
- seconds
- unexpected end of interval
sasha Updated translation strings. CLOSED
|
|
(14) [F] The same error message for incorrect "Update interval" and "Custom intervals".
For example, if "update interval" is empty, error message "Incorrect value for field "delay": invalid delay". And the same error, if in Custom interval "period" is empty.
iivs Now in API and DB there is only one field called "delay" that contains both delay and intervals separated by a semicolon. Visualy we didn't change this. Message comes from API as it should be. So...
WON'T FIX
|
|
(15) [D][I] Error "Incorrect value for field "lifetime": invalid lifetime" in LLD rule configuration when "{$M}d" is entered as Keep lost resources period. Database upgrade depends on this construction to work since macros were supported in this field before and there is no chance we can update macro value without breaking it's usage in other places.
glebs.ivanovskis It was decided to overwrite all macros used in this field with maximum allowed value during upgrade and document this in Upgrade notes.
glebs.ivanovskis RESOLVED in r65746.
vjaceslavs CLOSED
|
|
(16) [F] r65757: Reduced the amount of code in CSimpleIntervalParser class. Please review.
iivs CLOSED
|
|
(17) [F] r65766,r65767: Reduced the amount of code in CTimePeriodParser class. Please review.
iivs CLOSED
|
|
(18) [F] r65768,r65989: Fixed parsing of time periods by CTimePeriodsParser class. Updated unit tests.
iivs Removed unused variable in r66078, otherwise good.
CLOSED
|
|
(19) [F] can't use time unit in User profile → Messaging - Message timeout. Also not changed string Message timeout (seconds) → Message timeout
sasha RESOLVED in r65898 (string), r66118 (time suffix)
iivs With minor coding style fix in r66232
CLOSED
|
|
(20) [F] error message "Incorrect method "user.resetErrors"." when try to update user profile with wrong (e.g. empty) "Refresh" or "Auto-logout"
sasha RESOLVED in r66074
natalja.zabbix if wrong "Auto-logout" error:
sasha RESOLVED in r66190
iivs Another problem with autologout in users and profile pages:
Checkbox for "Auto-logout" returns to checked state when it was unchecked, but there is an error on page.
REOPENED
sasha RESOLVED in r66469
iivs CLOSED
|
|
(21) [F] r65809: Fixed parsing of scheduled intervals by CSchedulingIntervalParser class. Updated unit tests.
iivs Interval md1md2 is invalid. There cannot be another month day. Other time units are affected as well. Added unit tests (that currently fail) in r66098
REOPENED
sasha Thanks a lot! RESOLVED in r66103
iivs Great!
CLOSED
|
|
(22) [S] Host/template macros do not work in item update intervals and history/trend storage period. RESOLVED in r65870, r66223.
vjaceslavs CLOSED
|
|
(23) [F] We need to display minutes and months differently. Here I have 180s update interval an 90d history storage period:
Also frontend does not use week suffix at all. It shows 1m 5d for 5w. Also (I guess it's an old issue) it shows 1y for 360d - 365d and 52w. Personally, I would not use months and years as time units because of their ambiguity. For example, Zabbix year is neither astronomical nor calendar one on ~25% of occasions.
glebs.ivanovskis (31) will partially resolve the problem.
sasha It is duplicate of ZBX-4646
CLOSED
|
|
subissue (23) seems to be a duplicate of ZBX-4646
glebs.ivanovskis But at the time of ZBX-4646 only update intervals were displayed this way. And given that intervals come close to months relatively rarely the issue was minor. Now history and trend storage periods are displayed the same way and they are more likely to hit months and years.
<richlv> ...and also graph titles like (1h 3m) or (3m) - and possibly other places
|
|
(24) [F] r65880: Fixed parsing of scheduled intervals by CUpdateIntervalParser class. Updated unit tests.
iivs CLOSED
|
|
(25) [S] Probably LLD should validate intervals and history/trend storage period and either not update these fields with invalid values for existing items or fail to create new items with invalid settings. Although it will come with additional cost of user macro resolving.
glebs.ivanovskis We discussed it and decided that it is not 100% necessary to have such validation right now. So we decided to postpone this subissue.
MOVED to ZBXNEXT-3848.
CLOSED
|
|
(26) [F] r65902: Disabled default support of macros by new parsers.
iivs With minor fix in r66106
CLOSED
|
|
(27) [F] r65914: API validator improvements
iivs CLOSED
|
|
(28) [I] config.refresh_unsupported in data.tmpl must be changed from 600 to 10m
glebs.ivanovskis Duplicates (7). I haven't touched much of data.tmpl, but I keep that in mind.
glebs.ivanovskis As discussed, changed config.refresh_unsupported and config.work_period defaults in r66048, r66049, r66055, r66108. RESOLVED
glebs.ivanovskis Reworked DCsync_config() to use defaults from DB schema in r66090. Still RESOLVED
sasha ZBX_PROXY flag must be removed from dbupgrade scripts
REOPENED
glebs.ivanovskis RESOLVED in r66113.
sasha CLOSED
|
|
(29) [FI] Slide show delay should support time unit suffixes for consistency.
glebs.ivanovskis Database upgrade and schema updates in r66056, r66096. Partially RESOLVED
sasha [F] RESOLVED in r66110
iivs CLOSED
glebs.ivanovskis [I] is still RESOLVED.
vjaceslavs CLOSED
|
|
(30) [FIS] The syntax of multiple periods must prohibit trailing ';' (working hours, action condition time period, user media activity period). Database upgrade must delete trailing ';' from these fields.
glebs.ivanovskis Database patch added in r66067 and fixed by sasha in (35) r66076. Server already prohibits trailing semicolon. Partially RESOLVED
iivs Trailing semicolon RESOLVED by sasha in (18) [AF] r65989. Upgrade patch needs testing.
RESOLVED
glebs.ivanovskis One more fix in r66115. Still RESOLVED
vjaceslavs [IS] CLOSED, [F] still RESOLVED
sasha [AF] already tested by iivs under (18).
[AFIS] CLOSED
|
|
(31) [F] In Configuration->Templates/Hosts->Items update intervals and history/trend storage periods should be viewed as they are displayed in item edit form to avoid confusion. Subfilters should still use conversions.
sasha RESOLVED in r66077
natalja.zabbix In Configuration->Templates/Hosts->Items prototypes update intervals still converted to seconds.
REOPENED
sasha RESOLVED in r66306
iivs CLOSED
|
|
(32) [I] New limits in action and operation step durations require database upgrade to crop values exceeding new limits to the maximum allowed value.
glebs.ivanovskis RESOLVED in r66068.
vjaceslavs CLOSED
|
|
(33) [F] r65873: minor code improvements in CFlexibleIntervalParser() class
iivs CLOSED
|
|
(34) [AF] r65923, r65938, r65940, r65947, r65953, r65954, r65955, r66074: code improvements during code review
iivs Some improvements and fixes from my side in r66098, r66106, r66185, r66196, r66208, r66210, r66225, r66232, r66252, r66260, r66283, r66285, r66288, r66299, r66300. Please, review.
RESOLVED
sasha I accept all changes, except changes in r66283. These changes broke API tests and absolutely useless.
REOPENED
sasha RESOLVED in r67665
iivs Really sorry about that. Thanks for fixing this.
CLOSED
|
|
(35) [I] r66076: fixed SQL error in database upgrade patch
glebs.ivanovskis Thank you!
CLOSED
|
|
(36) Blank screen on login page after successful setup on a clean install of r66093 (PostgreSQL).
/var/log/apache2/error.log:
It complains about missing profiles, but everything seems to be fine in the database:
sasha Cannot reproduce. WON'T FIX
glebs.ivanovskis Issue resolved all by itself around r66108.
|
|
(37) [I] As discussed, upgrade patch should not change default filed values like 0 -> 0s, 0 -> 0d, etc. Upgrade patch should not change 0 to 0s or 0d, especially in those places where 0 is used as a special value (e.g. 0 - use default).
glebs.ivanovskis RESOLVED in r66108.
vjaceslavs CLOSED
|
|
(38) [F] r66114: removed useless ZBX_DEFAULT_OPERATION_STEP_DURATION
iivs Removed few more unused defines in r66916.
CLOSED
|
|
(39) [I] Upgraded PostgreSQL database has some differences:
items.delay, operations.esp_period and slides.delay
--- upg.dump 2017-03-02 17:50:34.665301640 +0200
+++ new.dump 2017-03-02 17:26:02.271114696 +0200
@@ -1237,7 +1237,7 @@
hostid bigint NOT NULL,
name character varying(255) DEFAULT ''::character varying NOT NULL,
key_ character varying(255) DEFAULT ''::character varying NOT NULL,
- delay character varying(1024) DEFAULT 0 NOT NULL,
+ delay character varying(1024) DEFAULT '0'::character varying NOT NULL,
history character varying(255) DEFAULT '90d'::character varying NOT NULL,
trends character varying(255) DEFAULT '365d'::character varying NOT NULL,
status integer DEFAULT 0 NOT NULL,
@@ -1474,7 +1474,7 @@
operationid bigint NOT NULL,
actionid bigint NOT NULL,
operationtype integer DEFAULT 0 NOT NULL,
- esc_period character varying(255) DEFAULT 0 NOT NULL,
+ esc_period character varying(255) DEFAULT '0'::character varying NOT NULL,
esc_step_from integer DEFAULT 1 NOT NULL,
esc_step_to integer DEFAULT 1 NOT NULL,
evaltype integer DEFAULT 0 NOT NULL,
@@ -1945,7 +1945,7 @@
slideshowid bigint NOT NULL,
screenid bigint NOT NULL,
step integer DEFAULT 0 NOT NULL,
- delay character varying(32) DEFAULT 0 NOT NULL
+ delay character varying(32) DEFAULT '0'::character varying NOT NULL
);
glebs.ivanovskis RESOLVED in r66119.
sasha To prevent such problems in future it must be fixed in DBmodify_field_type_sql() function
REOPENED
glebs.ivanovskis Please review r66120, r66131.
My only concern is that this approach leads to duplicate queries, e.g.:
Also removed redundant ;\n so that
becomes
RESOLVED
sasha Thanks! CLOSED
|
|
(40) [I] missed time unit 'd' for existing Template->Discovery rule in field "Keep lost resources period"
glebs.ivanovskis Related to (7). RESOLVED in r66209.
vjaceslavs CLOSED
|
|
(41) [F] Latest data shows nothing even though I can see data in graphs and screens.
Results for this query are not empty.
sasha RESOLVED in r66268
iivs CLOSED
|
|
(42) [AF] Mass update of update interval drops item preprocessing options. I have not checked in trunk, could be a common issue.
glebs.ivanovskis Same thing in trunk.
CLOSED and MOVED to ZBXNEXT-1443.(41)
|
|
(43) Updated to the latest trunk in r66242, conflicting files were:
- create/src/schema.tmpl
- frontends/php/include/defines.inc.php
- frontends/php/include/profiles.inc.php
- frontends/php/include/schema.inc.php
- frontends/php/tests/selenium/testFormAdministrationGeneralHousekeeper.php
- src/libs/zbxdbcache/dbconfig.c
- src/libs/zbxdbupgrade/dbupgrade_3030.c
- src/zabbix_server/escalator/escalator.c
- src/zabbix_server/housekeeper/housekeeper.c
- src/zabbix_server/httppoller/httptest.c
Please review.
iivs PHP code REVIEWED.
glebs.ivanovskis Another non-trivial merge from trunk in r67523. Resolved conflicts in:
- create/src/schema.tmpl
- frontends/php/include/classes/api/services/CHttpTest.php
- frontends/php/include/classes/validators/CApiInputValidator.php
- frontends/php/include/defines.inc.php
- frontends/php/tests/unit/include/classes/validators/CApiInputValidatorTest.php
- src/libs/zbxdbhigh/host.c
- src/libs/zbxdbupgrade/dbupgrade_3030.c
- src/zabbix_server/httppoller/httptest.c
- tests/data/data_test.sql
Please have a look, particularly at CHttpTest.php.
sasha Excellent! CLOSED
|
|
(44) [S] Server crash. Probably related to update in r66242.
glebs.ivanovskis Root cause: ZBX-11829.(7)
WON'T FIX
|
|
(45) [F] Sorting by intervals/history/trend doesn't work correctly with time units
Steps to reproduce:
1. Create items with update intervals 2m, 10m, 3h, 30s
2. Go to Configuration->Templates/Hosts->Items
3. Sort items by intervals ASC
Expected: 30s, 2m, 10m, 3h
Actual: 2m, 3h, 10m, 30s
iivs Another problem related to sorting. For example, trapper item types have no delay and it's displayed empty. The sorting by delay is now random. Those trapper items are somewhere in the middle after "0", "0d", "0m" but before "0s", "0w". This affects items, item prototypes and LLD rules. Sorting by trends is also affected.
sasha RESOLVED in r66306, r66341 and r66342
iivs Sorting by trends, empty values may still appear somewhere in the middle.
REOPENED
sasha RESLOVED in r67265
iivs CLOSED
|
|
(46) [F] In latest data with enabled show details and enabled housekeeping, the values for history and trends are shown in seconds.
sasha RESOLVED in r66341
iivs CLOSED
|
|
(47) [F] Latest data (latest.php):
- We resolve macros in line 223, but later on in code we use parser with macro option enabled. In case macros are not resolved, it still will show the value as was entered. So there is no need for that option.
- Non-integer item values types show trends with value 0. It used to show empty space.
With housekeeping enabled, history and trends are shown in seconds. Duplicate of (46).
sasha RESOLVED in r66306, r66341, r66342 and r66397
iivs CLOSED
|
|
(48) [AF] Maybe let's trim leading zeroes like we used to before and save them in DB. 000 and 0 means the same thing. Also we have suffixes like 0d, 0m, 0w that also mean the same thing. Thus searching by 0, nothing is found and the sorting is also alphabetical.
sasha RESOLVED in r66484
iivs CLOSED
|
|
(49) [F] Searching by delay has become hard if we have flexible or scheduling intervals. If delay is 0, it's not possible to seach by 0. Also if delay is any other value, since it's now one field and we search the exact match.
sasha RESOLVED in r66483
iivs CLOSED
|
|
(50) [F] In edit form the history and trends (if enabled) consolidates the value 69d -> 2m 9d. We could display the value only in days, by extending the conversion function with additional parameter to display the desired output.
sasha RESOLVED in r66424
iivs CLOSED
|
|
(51) [F] T_ZBX_PERIOD and T_ZBX_PERIOD_MACRO is actually interval and can be mistaken with time period. Let's rename it to time unit like T_ZBX_TU, for example.
sasha RESOLVED in r66426, r66429
iivs CLOSED
|
|
(52) [F] Several places use strpos(), for example in forms.inc.php L684, but we already know that it's a string and cannot contain macros in the middle of the string. So we can change (strpos($trends, '{') === false) to $trends[0] === '{'. Please, check other places as well.
sasha RESOLVED in r66397
iivs CLOSED
|
|
(53) [F] Status of Zabbix available in screens for all users and it's the same widget used in other places. We could show that widget in reports page and dashboard now too.
sasha Let's make it under more related task! Moved to ZBXNEXT-2102 (1).
WON'T FIX
iivs Good idea.
CLOSED
|
|
(54) [F] In items, item prototypes, discovery that uses custom intervals, we could use additional validation in check_fields() for "delay", since we have a separate temporary parameter "delay_flex". That way we can no longer enter full update interval in small delay field.
sasha RESOLVED in r66429
iivs CLOSED
|
|
(55) [F] Replace hardcoded 2147483647 with something different.
sasha RESOLVED in r66415
iivs CLOSED
|
|
(56) [A] CItemGeneral.php L285: That causes a new bug. When scheduling interval exists, but all other delays (also flexible is given) are 0, the string passes validation.
sasha RESOLVED in r66422
iivs CLOSED
|
|
(57) [F]
- discoveryconf.php L259: Wrong default value used "httptest".
- slideconf.php
- L35: Translation string says "in seconds". That is no longer true.
- L35: BETWEEN(1, SEC_PER_DAY) is not properly validated by check_fields(), it gives two different error messages when validating same time unit.
Incorrect value "604800" for "Default delay (in seconds)" field: must be between 1 and 86400.
and
Incorrect value for field "delay": must be between "1" and "86400".
- jsrpc.php
- L84: Could cause problems, although somehow the messaging still works.
- L174: Could use CWebUser::getSessionCookie().
- administration.users.edit.php: L162 $data vs L163,169 $this->data.
sasha RESOLVED in r66434
iivs CLOSED
|
|
(59) [F] SQL error in problem view (Monitoring -> Problems):
vjaceslavs Looks like there was a was some problem with a DB migration from previous tests. Recreated DB, problem is gone, so
CLOSED
|
|
(64) [AF] Next check of web scenarios and network discoveries is not being reset on changes of update interval.
iivs RESOLVED in r67619
sasha Successfully tested! Have a look at my changes in r67666.
iivs CLOSED
|
|
(65) [AF] Range checking is broken in web scenario step timeout and user profile refresh and auto-logout settings. It cannot be fully explained by my stupidity in r67523, because I touched only web scenario step timeout validation there.
iivs It was due to incorrect define. RESOLVED in r67568
sasha CLOSED
|
|
(66) [D] Related to (64). Resetting nextcheck to 0 will help in case of network discovery and web scenario update interval changes like 1w->1h. However, if a user macro is used in update interval and its value changes 1w->1h, nextcheck will remain far in the future. Proper fix would involve moving discovery rules and web scenarios into server's configuration cache and will not happen within this task. For now it should be documented as a limitation.
martins-v Added to:
RESOLVED
glebs.ivanovskis Thank you!
CLOSED
|
|
(67) [PS] Another error message wasn't properly formatted. RESOLVED in r67618.
vjaceslavs CLOSED
|
|
(68) [PS] Handling of invalid intervals in web scenarios and network discovery was not ideal. Server used to set next check to year 2038 with no way back. Now it will use Refresh notsupported items setting to schedule next check of "unschedulable" network discovery or web scenarios (without actually executing them). This should both give user a chance to fix configuration and protect server from excessive load. RESOLVED in r67667.
vjaceslavs CLOSED
|
|
(69) [S] Server complains about 0 in global history/trends storage period override:
Thanks to vjaceslavs for spotting missing '*'! RESOLVED in 67806.
vjaceslavs CLOSED
|
|
(71) [FS] SQL errors in Administration->Proxies, frontend tries to calculate required proxy performance like that:
glebs.ivanovskis Server will now provide necessary per-proxy information within status.get protocol. [S] RESOLVED in r67938, r67944, r67954, r67955, r67959, r67960, r67966, r67971.
sasha [F] RESOLVED in r67974
oleg.egorov [F] CLOSED
glebs.ivanovskis [S] now properly RESOLVED after r68087, r68101, r68102, r68109.
Note that there were two significant changes in behavoiur: LLD rules are no longer counted in required performance and web scenario items will be now counted along with other types of items. And we now count all items on disabled hosts as disabled.
glebs.ivanovskis Stylistic improvements in r68199. [S] is still RESOLVED
vjaceslavs Looks good to me! CLOSED
|
|
(72) [P] Proxy is crashing when syncing configuration from the empty database for the first time. Problem is that if there are no rows in config table we try to take data from schema defaults and there is no default for discovery_groupid. RESOLVED in r67829.
vjaceslavs CLOSED
|
|
(73) [PS] Please have a look at my minor changes in r68097 and r68096.
sasha I don't see sense in these changes. Please describe it!
+#if USER_TYPE_UNDEFINED != 0
+# error "expression below assumes that atoi() of non-numeric value will yield USER_TYPE_UNDEFINED too"
+#endif
glebs.ivanovskis Well, maybe I went too paranoid. Reverted r68097.
vjaceslavs CLOSED
|
|
(74) [D] Status of Zabbix and Administration->Proxies descriptions will need an update after (71) is settled down. Seems no need to update internal item documentation and no need to mention this in upgrade notes.
glebs.ivanovskis Please review my changes to the aforementioned pages. RESOLVED
martins-v Looking good to me. I know that sasha wanted to have a look, too.
glebs.ivanovskis He had and he wanted to drop one sentence in Number of items:
Monitored and not supported items are counted only on enabled hosts. Items on disabled hosts are counted as disabled.
I have no strong objections against it. So it's up to you.
martins-v I've removed the sentence.
glebs.ivanovskis CLOSED
|
|
Available in pre-3.4.0alpha1 (trunk) r68215.
|
|
(75) [S] Defect reported by Coverity (CID 162508). Invalid user media activity period is not handled properly.
glebs.ivanovskis RESOLVED in a new svn://svn.zabbix.com/branches/dev/ZBXNEXT-1675 r68223.
vjaceslavs CLOSED
|
|
Fix for (75) is available in pre-3.4.0alpha1 (trunk) r68228.
|
|
Fixed in:
- 3.4.0alpha1 (trunk) r68675
|
discovered_item_editing.png
Team A
