-
-
Notifications
You must be signed in to change notification settings - Fork 439
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
"End of system activity file unexpected" during sadf #173
Comments
Hi, Note: Please send me your binary sar file to sysstat at orange dot fr (I don't have access to Google drive). |
Hi there, Thanks a lot for looking into this! Sorry that I forgot to grant permission on the above sar binary file,it should be fixed now. I also sent a copy of the broken binary sar_log to the email address you provided. Is the sadc collector not writing data to disk in an atomic fashion? What's the proper termination signal we should send in this case then? Thanks, |
Sorry for my previous answer. In fact it can't be because of the SIGINT signal interrupting sadc since this signal is trapped by sadc.
|
Thanks for the reply! We create it by ssh into the machine and run the command i posted above, and it's running on Hardware. We just got another sar binary: https://drive.google.com/file/d/1tIvaVSqEOgol5BokQQUCI9Jq6uzXOqI0/view?usp=sharing Here's the part where it starts to become crazy from your sar -f command:
Which file under /proc do we get the cpuinfo from? Maybe we can have a separate thread to ensure we have access to it while running our sar command... |
Another interesting observation is that the time in the /history/log/sar_log.stdout (from the sar command stdout) and
The output from
|
Recently I also hit the same issue on my system and it seems critical as we are unable to analyze the performance statistics. When I tried extracting one of the sa content, it initially displayed the stats from 12:00:01 to 12:24:11 12:00:01 AM CPU %user %nice %system %iowait %steal %idle However after ~24 mins, the timer seems to clock again from 12:00:00 with absolutely zero statistics and this continues for few secs logging huge redundant statistics to the sa file eventually ending with the message "End of system activity file unexpected": 12:00:00 AM all 0.00 0.00 0.00 0.00 0.00 0.00 It seems like sysstat has corrupted the file due to some unknown reason. I see this defect is still open. Do we have plan for fix. What can be done to avoid this issue if it occurs more frequently? FYI, I am running sysstat 10.1.5 version on centos7.4 OS. Will upgrading systat version helps? |
I cannot promise that upgrading will fix the problem since I have been unable to reproduce it and to give any proper explanations. |
Hi, We hit this issue several times. I am not able to reproduce but here are some hints. It obviously happens here: Line 1404 in 6886152
This method being called at several places, it is difficult for me to say where exactly it happens. In my case, it occurs after power-cycling and normally, sar should print a line about LINUX RESTART. Something like:
This is not the case so I would assume something happens at this moment.
I played with Restarting the service helps. The unit contains only the execution of Hope this can give some hints. |
@aquemy Please also send me your saXX binary data file so that I can analyze it (send it to sysstat at orange dot fr). |
@akhishukla23 FYI I have been reported that same problem ("End of system activity file unexpected") on CentOS 7.7 using sysstat 10.1.5. Altough I'm pretty sure that this issue is related to a lack of disk space, it seems to have been fixed by disabling a cron job calling |
@sysstat Earlier you'd asked if the log was created in a vm. What is the significance of that? I am now seeing that all of the sar logs that I create with vms are reporting "End of system activity file unexpected" where all of the logs on physical machines which are collected the same way are working fine. |
I am using centos 7.7 with sysstat 10.1.5 and I see this issue on all virtual machines constantly. Is there an issue I can track somewhere please? |
@sysstat
The following changes were made to this
and saxx file is created in the /var/log/sa/201912/ path. But, Problems arise when executing.
An error occurs as you move on to the next day. |
@sysstat I'd love to know if there is any active investigation regarding the issue where sar data files are unreadable for some metrics (e.g -n ETCP) on Virtual Machines. |
@gleventhal Sorry, but as I have already told before, I have never been able to reproduce the problem on my machine. Moreover I don't' use VM (and I really don't have the time to set up such an environment). |
@sysstat For details about how to reproduce the problem, see this link.:https://bugzilla.redhat.com/show_bug.cgi?id=1669932 |
@tong-da-wei Use |
I have the same problem,Can you help answer me @sysstat |
@mrzhangxing
|
I am also having this problem on Centos 7.9, with sysstat version 10.1.5.
Then I had the issue of
Still had the same issue. Solved it by uninstalling sysstat package, downloading the latest code and building the latest version, which worked. |
Here're the steps we did to hit this error:
We are using:
sysstat version 11.2.0
(C) Sebastien Godard (sysstat orange.fr)
Step 3) above hits the "End of system activity file unexpected" error and we have checked that our /history partition has space. The binary sar file can be found here:https://drive.google.com/file/d/1hlYOSwPyqv3uI-BJrrIazaXvZavGXV0d/view?usp=sharing
Please let us know if anything else may be useful to diagnose the problem and whether this is a known bug somewhere.
The text was updated successfully, but these errors were encountered: