[ZBX-7649] icmp ping items for a single host should be checked together Created: 2014 Jan 14 Updated: 2017 May 30 Resolved: 2014 Jan 23 |
|
| Status: | Closed |
| Project: | ZABBIX BUGS AND ISSUES |
| Component/s: | Proxy (P), Server (S) |
| Affects Version/s: | 2.2.1 |
| Fix Version/s: | 2.2.2rc1, 2.3.0 |
| Type: | Incident report | Priority: | Minor |
| Reporter: | Aleksandrs Saveljevs | Assignee: | Unassigned |
| Resolution: | Fixed | Votes: | 0 |
| Labels: | icmpping | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Description |
|
We have "Template ICMP Ping" in the default installation of Zabbix with three items: icmpping, icmppingsec, and icmppingloss. The time these items will be checked is calculated based on their item ID, so most probably these items will be scheduled for a check at times T, T+1, and T+2. Now, suppose Zabbix server only has one pinger running. It will take the first item at time T and process it. Processing will take around 3 seconds, so the next time it looks into the queue it is going to process the other two items. So in order to obtain values for a single host two passes are needed. It would be nice to check all three items with a single fping invocation. The easiest way of doing this is scheduling a check based on their interface ID, rather than on item ID. |
| Comments |
| Comment by Andris Zeila [ 2014 Jan 15 ] |
|
Fixed in development branch svn://svn.zabbix.com/branches/dev/ZBX-7649 |
| Comment by Oleksii Zagorskyi [ 2014 Jan 15 ] |
|
(1) Would be nice to reflect the changes a bit on: wiper Added scheduling optimization information. Please review. RESOLVED. asaveljevs Looks good. I only fixed a typo in the second link. CLOSED. |
| Comment by Andris Zeila [ 2014 Jan 22 ] |
|
(2) There is no need to base java/pinger poller item nextchek seed calculations on interfaceid and poller type as they already have different interfaces. RESOLVED in r41727 asaveljevs CLOSED |
| Comment by Aleksandrs Saveljevs [ 2014 Jan 22 ] |
|
(3) In the email, sasha offered to rename __config_pinger_nextcheck_compare() and other similar functions to be consistent. You seem to have agreed, but decided not to do it. Reminding about it in this subissue. However, it might not be worth doing it, because I am changing this code a bit in wiper Right, forgot about it. RESOLVED in r41762 asaveljevs Yes, let's revert that. Did it in r41780. CLOSED. |
| Comment by Aleksandrs Saveljevs [ 2014 Jan 22 ] |
|
(4) In DCsync_items(), around line 960, we set poller_type to ZBX_NO_POLLER, and then calculate nextcheck seed. Since get_item_nextcheck_seed() relies on poller_type, it would be nice to calculate the seed after poller_type is set. wiper RESOLVED in r41760 asaveljevs CLOSED. |
| Comment by Aleksandrs Saveljevs [ 2014 Jan 22 ] |
|
(5) Let's adapt function calculateItemNextcheck() in the frontend, too! asaveljevs Adapted in r41785. RESOLVED. wiper CLOSED |
| Comment by Aleksandrs Saveljevs [ 2014 Jan 22 ] |
|
(6) Please take a look at minor fixes in r41758. wiper thanks, CLOSED |
| Comment by Andris Zeila [ 2014 Jan 22 ] |
|
Subissues (3), (4) and (6) are fixed. Did not set to resolved state because of (5) |
| Comment by Aleksandrs Saveljevs [ 2014 Jan 23 ] |
|
(7) Note that after the changes JMX items will be scheduled based on interface ID if their host is reachable, and based on item ID if not (because it will then be processed by an unreachable poller). This is probably not a problem, because this change in scheduling will only affect items whose delay settings change between calls to DCsync_items(). During regular processing, JMX items for unreachable hosts will still be scheduled based on host->jmx_disable_until, as returned by DCget_unreachable_nextcheck() function. wiper CLOSED |
| Comment by Andris Zeila [ 2014 Jan 23 ] |
|
Released in: |