You are browsing a read-only backup copy of Wikitech. The primary site can be found at wikitech.wikimedia.org

User talk:Paucabot

From Wikitech-static
Jump to navigation Jump to search

Welcome to Toolforge!

Hello Paucabot, welcome to the Toolforge project! Your request for access was processed, and you should be able to use ssh to connect to login.tools.wmflabs.org. You will need to logout and login again at https://toolsadmin.wikimedia.org/ to activate your new permissions there.

Check the Toolforge help page for tips on using your account. You can also ask questions in our IRC channel at #wikimedia-cloud connect or send an e-mail to our mailing list cloud@lists.wikimedia.org.

Thank you, and have fun making Tools! --StrikerBot (talk) 21:04, 2 January 2019 (UTC)


Your cronjob problem reported on irc 2020-05-09

Sorry nobody noticed your question while you were still in the channel. I took a look at your rebot tool and think I figured out the problem. There was a job stuck in Error state on the job grid. Here's how I found that and what I did to try and fix it for you:

The can't get password entry for user "tools.rebot" error reason is something that happens occasionally to grid jobs. It usually means that some error has happened on the grid engine scheduler node which prevented it from looking up your tool's unix account information in our LDAP directory. These are unfortunate, but transient errors. You should have been sent an email by the grid engine service when this happened, but those emails are sometimes sent to spam folders or even rejected entirely by some mail servers.

Anyway, with the broken job deleted I would expect your bot's job to run starting at 13:43 on 2020-05-10. --BryanDavis (talk) 20:16, 9 May 2020 (UTC)

Thanks for your detailed information and for the solution, BryanDavis. I'm so grateful! Thanks again! Paucabot (talk) 20:39, 9 May 2020 (UTC)
Hi BryanDavis. My cron job seems to be stuck again. I tried to do what you told me, but this time it seems there is no job in qstat. What am I doing wrong? Thanks in advance, Paucabot (talk) 17:52, 12 May 2020 (UTC)
$ sudo become rebot
$ crontab -l
# Wikimedia Toolforge specific note:
#   Please be aware that *only* jsub and jstart are acceptable
#   commands to schedule via cron.  Any command specified here will
#   be modified to be invoked through jsub unless it is one of
#   the two.
#
# m     h       dom     mon     dow     command
0 22 * * * /usr/bin/jsub -N cron-29 -once -quiet screen - m -d sh robot.sh
This crontab line is attempting to start a GNU Screen instance on the job grid. Looking at $HOME/cron-29.out I can see that this is failing with the message "Must be connected to a terminal". Cron is starting the job, but the job itself is finishing almost immediately with this error. This is not surprising to me as Screen is a terminal multiplexer application intended for interactive use rather than running as a workload on a distributed job grid. It is also designed not to terminate, so if you did manage to get it running on the grid you would be back to the error of having an active job of the same name on the grid error from before, although for a slightly different reason.
I think your crontab line should look more like 0 22 * * * /usr/bin/jsub -N cron-29 -once -quiet /data/project/rebot/robot.sh. You might want to change -N cron-29 to a more descriptive name for the job. --BryanDavis (talk) 23:01, 12 May 2020 (UTC)
Thanks, BryanDavis. As you can see, I am a newbie here and I maybe lack the he knowledge to do some things. I'm very grateful to you for your patience and explanations.
I have changed my crontab accordingly to what you said. The reason why I introduced the screen command was because I was not sure how to access to the job and stop it, but with your answers, you helped me (I learned about qstat) and I think now I will be able to stop the bot if something goes south.
Thanks again for your help. Paucabot (talk) 06:09, 13 May 2020 (UTC)