[ZBX-13769] inconsistent snmpV3 host availability detection in case of wrong credential parameters Created: 2018 Apr 19 Updated: 2024 Apr 10 Resolved: 2024 Jan 05 |
|
Status: | Closed |
Project: | ZABBIX BUGS AND ISSUES |
Component/s: | Documentation (D) |
Affects Version/s: | 3.0.16, 3.4.8 |
Fix Version/s: | 4.0 (plan) |
Type: | Problem report | Priority: | Trivial |
Reporter: | Oleksii Zagorskyi | Assignee: | Martins Valkovskis |
Resolution: | Duplicate | Votes: | 5 |
Labels: | availability, consistency, credentials, notsupported, snmpv3 | ||
Remaining Estimate: | Not Specified | ||
Time Spent: | Not Specified | ||
Original Estimate: | Not Specified |
Issue Links: |
|
||||||||||||||||
Team: | |||||||||||||||||
Sprint: | Sprint 32, Sprint 33, Sprint 34 |
Description |
For SNMP v3 items we have 5 parameters to configure authentication (we asume we use AuthPriv mode):
Imagine that some of these parametr(s) is configured incorrectly. Here are conclusions based on tests performed in command line and on zabbix server (restarted each time after changes in frontend because of 1. Wrong "Security name" -u causes item became unsupported 2. Incorect auth -a OR wrong -A value causes item becoming unsupported 3. Incorrect mode -x causes SNMP agent/interface availability 4. Wrong "Privacy passphrase" -X causes real timeout and SNMP agent/interface availability Conclusion:
What would be correct to fix for these two - really hard to say, should be discussed. I personally, after this investigation, cannot select the optimal answer. This is highly related to the mentioned |
Comments |
Comment by Oleksii Zagorskyi [ 2018 Apr 20 ] |
|
Comment by Oleksii Zagorskyi [ 2018 Apr 23 ] |
One more thing - as for SNMP agent I used a Linux box with net-smp v5.7 for snmpd daemon. Original issue was discovered on Cisco F5 device. |
Comment by Dimitri Bellini [ 2018 Jul 13 ] |
I have noticed the same behaviour on Nexans switch. From the Zabbix Server log i can read: item "Switch5:system.uptime" became not supported: Cannot connect to "xxx.xxx.xxx.xxx:161": Unknown user name. But the Zabbix availability icon is green.... I think we need to put the icon on "RED" |
Comment by Oleksii Zagorskyi [ 2019 Nov 13 ] |
There is mistake in documentation change. I'd write the sentence this way:
REOPENED |
Comment by Ilya Ableev [ 2020 Jul 16 ] |
Up? |
Comment by Oleg Ivanivskyi [ 2022 May 02 ] |
5. Too "simple" authentication passphrase (-X) may cause "network errors"/timeouts and flapping SNMP agent availability for Cisco. Had an issue with "flapping" SNMP agent availability on hundreds Cisco switches and routers (red > green > red > ...). All devices were configured with a simple "1234567890" auth passphrase for SNMP v3 user (e.g. "-l authPriv -u TestUser1 -a SHA -A Example@22 -x AES -X 1234567890"). I was able to poll devices via SNMP from CLI (no errors at all). At the same time, Zabbix 4.4 generated many "network errors" in the log continuously for "random" SNMP items. All of device performance graphs had gups. This issue was fixed by changing "1234567890" auth passphrase to a more complex one (e.g. "AuthPass1!"). SNMP status isn't flapping any more. No gaps on the graphs. |
Comment by Oleksii Zagorskyi [ 2022 May 02 ] |
Oleg, I believe that "1234567890" is absolutely fine as a value for auth/encryption. I do not see any possible scenario why it might cause availability flapping. |
Comment by Oleg Ivanivskyi [ 2022 May 02 ] |
Note, I am talking about ~700 Cisco devices. I checked the EngineID for all of them. There were no duplicates. |
Comment by Richard Ostrochovský [ 2023 Sep 27 ] |
Isn't this related also to |
Comment by Oleksii Zagorskyi [ 2024 Jan 05 ] |
Looks like the So, this should be closed as Duplicate. Done. |