Some 1.3.4 bugs

[eran]

Hello,
I'm new to Zabbix so I can't say if any of these bugs are new or existed in previous versions. I choose to install the latest beta because I need the web monitoring capabilities which are not available in the stable version.

I will begin with what I consider the more serious bugs:

1. Windows eventlog monitoring simply doesn't work. I configured two eventlogs to be monitored ("Application" and a custom log called "Newman"), none of the events in either of these logs are collected. Checking with a sniffer I can see that when the agent starts it's requesting the list of active checks and receives the two log keys. After that the agent doesn't send any more data to the server and the agent log file doesn't show anything about the eventlog or active checks.
   I tried this with the agents from version 1.3.4 and from 1.1.7, both on Windows 2003 R2 servers.
2. Long performance counter names are trimmed when the server sends them to the client. For example, I have this performance counter defined:
   perf_counter[\MSMQ Queue(web101\private$\mailprocessingqueue)\Message s in Queue]
   In the agent log file I see the request has been trimmed to [perf_counter[\MSMQ Queue(web101\private$\mailprocessingqueue)\Me] which of course results in [ZBX_NOTSUPPORTED].
   It will also be very helpful if the built-in macros like {HOSTNAME} could be used in keys so I could create the item once, in the template, and it will replaced with the host name. Something like this:
   perf_counter[\MSMQ Queue({HOSTNAME}\private$\mailprocessingqueue)\Mes sages in Queue]
3. Host status tends to get stuck as "unreachable", although other data is successfully received from the agent. When is say "tends" I mean that about 80% of the time my monitored servers are listed as "unreachable", once in a while a server would become reachable again and then go back to unreachable. I want to stress again that during this "unreachability" time, I get other readings from the server without any problem.
   This happens for both linux and windows agents.
4. CPU times on Windows agent are completely off.
   I'm not sure how they are measured but the numbers don't seem related to the numbers I see in performance monitor. I also get numbers like 3.0 which, to the best of my understand means 300%.

Now to the minor bugs:

1. When changing color in a graph to a number that has hexadecimal digits (a-f) there is an error "Warning. Field [color] in not integer".
2. The web monitoring tab remembers the filter settings of the previous page but shows the filter as "group-all", "host-all"
3. There is no way to logout from the system.
4. When there is only one screen saved changing the time settings on that page results in an error: "Critical error. Field "screenid] is not integer". When reviewing the URL I can see that the screenid parameter is empty. Adding another saved screen solves this problem.

It's quite a long list - I hope no one minds I didn't put them all to separate posts.

Eran

[Alexei]

Thanks for reporting these problem. I updated list of known issues at http://www.zabbix.com/wiki/doku.php?id=bugs:1.3.x