inode test fails after change in showgraph
Hello,
Yesterday Xymon server, version 4.3.12, stopped performing the test inode, after a change in program showgraph.cgi.
In all Xymon pages, we show the date in an ISO8601 format. File showgraph.c
is modified to generate an ISO8601 date too, by changing the format in the
invocations of strftime. After a make, a make install and a restart of
xymon server, the dates in the graphs were changed to the desired format.
However, after half an hour, all inode tests became purple. It was found that:
- in subdir ~xymon/server/bin only programs showgraph.cgi and xymond_rrd are rebuild, all the other programs have modification time multiple months back;
- there are no inode-related (error)messages in any of the logfiles in /var/log/xymon
make installhas changed some of the *.cfg files in ~xymon/server/etc. Those changes are reverted, by reloading the previous version of those files from an svn repository. This had no effect on the colour of the inode tests;- defining the INODE tag in ~xymon/server/etc/analysis.cfg for one of the affected hosts does not change the colour of the inode-test of that host. The colour of the test stays purple;
- reloading programs showgraph.cgi and xymon_rrd from a back-up does not change the colour of the inode-tests either. Of course the dates in the graphs are now again shown using the original format.
Xymon server has been stopped and started at each change. The configuration of the xymon clients has not been changed while looking for a solution for this problem.
Does any one knows how to solve this problem? Are there more things I can do (check,restore) to get the inode test operational again?
Regards, Wim Nelis.
The NLR disclaimer is valid for NLR e-mail messages.
This message is only meant for providing information. Nothing in this e-mail message amounts to a contractual or legal commitment on the part of the sender. This message may contain information that is not intended for you. If you are not the addressee or if this message was sent to you by mistake, you are requested to inform the sender and delete the message. Sender accepts no liability for damage of any kind resulting from the risks inherent in the electronic transmission of messages.
Den 09.01.2014 07:55, W.J.M. Nelis skrev:
Hello,
Yesterday Xymon server, version 4.3.12, stopped performing the test inode, after a change in program showgraph.cgi. [...] Are there more things I can do (check,restore) to get the inode test operational again?
If you look at the client data (follow link from any detailed status page), is there an "[inode]" section listed there with the necessary data?
It is the xymond_client module that generates the inode status, not xymond_rrd or showgraph.cgi. So if that binary is unchanged, then I don't really understand why it would stop generating an inode status.
Are there any notices in xymond.log (or the xymond status page) of status messages being dropped, e.g. from ghost clients?
I suppose the other status messages (cpu, disk, memory etc) are still working fine?
Regards, Henrik
Hello Henrik,
Den 09.01.2014 07:55, W.J.M. Nelis skrev:
Hello,
Yesterday Xymon server, version 4.3.12, stopped performing the test inode, after a change in program showgraph.cgi. [...] Are there more things I can do (check,restore) to get the inode test operational again?
If you look at the client data (follow link from any detailed status page), is there an "[inode]" section listed there with the necessary data? Yes, I forgot to mention that in the previous e-mail. However, this is *not* true for one server, the xymon server itself! All other clients still do send data, including a [inode] section. None of the modules in ~xymon/client/bin is changed yesterday, but the configuration files in ~xymon/client/etc have been changed. (As it took two runs to get the showgraph-patch working, the *.cfg.bak files are identical to *.cfg files.)
It is the xymond_client module that generates the inode status, not xymond_rrd or showgraph.cgi. So if that binary is unchanged, then I don't really understand why it would stop generating an inode status.
Are there any notices in xymond.log (or the xymond status page) of status messages being dropped, e.g. from ghost clients?
No, there are no messages about dropped status messages in xymond.log. The history of xymond status contains only messages about unrelated problems.
I suppose the other status messages (cpu, disk, memory etc) are still working fine?
Yes, those tests are working fine.
Regards, Wim Nelis.
The NLR disclaimer is valid for NLR e-mail messages.
This message is only meant for providing information. Nothing in this e-mail message amounts to a contractual or legal commitment on the part of the sender. This message may contain information that is not intended for you. If you are not the addressee or if this message was sent to you by mistake, you are requested to inform the sender and delete the message. Sender accepts no liability for damage of any kind resulting from the risks inherent in the electronic transmission of messages.
Den 09.01.2014 10:12, W.J.M. Nelis skrev:
It is the xymond_client module that generates the inode status, not xymond_rrd or showgraph.cgi. So if that binary is unchanged, then I don't really understand why it would stop generating an inode status.
Are there any notices in xymond.log (or the xymond status page) of status messages being dropped, e.g. from ghost clients? No, there are no messages about dropped status messages in xymond.log. The history of xymond status contains only messages about unrelated problems.
I suppose the other status messages (cpu, disk, memory etc) are still working fine?
Yes, those tests are working fine.
I wonder if inode checks has somehow been disabled. If you go to where you compiled Xymon, then in the xymon-4.3.X/lib directory there is a "loadhosts" binary. This is a debug module. Could you try running
cd xymon-4.3.12/lib
./loadhosts @ <hostname>
Pick item: XMH_HOSTNAME
... should show the hostname ...
Pick item: XMH_NOCOLUMNS
... should say 'XMH_NOCOLUMNS not found' ...
If it lists "inode" in XMH_NOCOLUMNS, then that is the problem.
If not, then try doing "kill -USR2 <pid of xymond_client>" - this toggles the debugging flag in xymond_client. It will start logging a lot of output to clientdata.log - does it show any messages with "Inode check" ?
You can turn off the debugging with the same kill-command.
Regards, Henrik
Hello Henrik,
the problem has been solved, that is the inode-test is running again. The log of the backup-service shows that all files in ~xymon/server/bin have changed on 2014.01.08, while only two of them (showgraph.cgi and xymon_rrd) had a modification time on that date. After restoring all files in subdirectory ~xymon/server/bin, the inode tests all became green.
The logs of the back-up service also show that the modification time of all files in ~xymon/server/bin have changed. The original times were on 2013.09.10, the date that Xymon 4.3.12 was installed, and the new times are on 2012.02.16 except for the two files already mentioned. I have no explanation for those weird modification times.
When compiling the new version of module showgraph, I assumed that the
make-files are still correct and do not need to be regenerated. I will
check if this assumption is correct by installing Xymon again, starting
with configure.
Thanks for your help, Wim Nelis.
The NLR disclaimer is valid for NLR e-mail messages.
This message is only meant for providing information. Nothing in this e-mail message amounts to a contractual or legal commitment on the part of the sender. This message may contain information that is not intended for you. If you are not the addressee or if this message was sent to you by mistake, you are requested to inform the sender and delete the message. Sender accepts no liability for damage of any kind resulting from the risks inherent in the electronic transmission of messages.
participants (2)
-
henrik@hswn.dk
-
Wim.Nelis@nlr.nl