Welcome to Tool Labs
Hello Beetstra, welcome to the Tool Labs project! Your request for access was processed and you should be able to login now via ssh at tools-login.wmflabs.org. To get started, check the help page and the migrating from the Toolserver page. You can also ask in our irc channel at #wikimedia-labs connect or send an e-mail to our mailing list email@example.com - thank you, and have fun using labs! petrb (talk) 10:58, 23 June 2013 (UTC)
I've had to stop linkwatcher, it was misbehaving for two reasons:
- It demonizes (fork and exit;) which prevents gridengine from managing them properly (and shouldn't be done in Tool Labs); and
- it runs under your user account, which should not be done for anything that runs unattended.
Please create tool accounts to run your tools; but also please revise the code so that they do not fork and run in the background like you are doing now. — Coren/Marc (talk) 14:48, 27 January 2014 (UTC)
- Hi Coren,
- I don't understand the first point - the bot runs, and it spawns numerous processes in parallel to do work for it. These processes 'fork-and-exit'. Those bots have been running like that for months on the grid without any problems. Can you be a bit more specific on this? (If the sub-processes don't fork-and-exit, they have to finish their job and can't run in parallel, if I put all the processes in the main bot the main bot has to do everything in which case it can not do multiple actions in parallel either .. )
- I will work on the second part - the last time that I tried that I could not get the files being owned by the <tool>-account ('tools-<toolname>').
- Thanks. --beetstra (talk) 04:26, 28 January 2014 (UTC)
- I managed to do the second part (files are however 'owned' by 'beetstra' in the project directory, but I can run them under my <toolname> account). --beetstra (talk) 04:44, 28 January 2014 (UTC)
- I have restarted the linkwatcher under an own account - I did not fix the fork-and-exit part (as I said, the grid-engine did not complain for months running this system, and IMHO it is the only way to be flexible and keep the very high speed with substantial amount of work that is needed for this). Please tell me if there are things that are intrinsically wrong with the system, but I think that if I jsub with #G of memory, and the bot and all the processes it needs to spawn stay within that, then there should not be a problem. (Or I really don't understand the problem). The cross-wiki anti-spam effort is greatly relying on these bots (linkwatcher/coibot/xlinkbot), and though there is no real damage, taking the bots down is hindering the anti-spam efforts greatly. If you don't mind, since the bots were running smoothly on the grid for months, can we leave them running while discussing/considering alternatives?
- I now get a problem with files that are shared between the bots ('coibot' is generating files that should be read by 'linkwatcher' and vice-versa - the rights are obviously not allowing that), I'll try and talk to people on IRC about that, but if you can point me to a 'manual' on that, that would be appreciated. Another question, is there a way to 'chown' files to the toolname, some datafiles (settings) need to be read, and the only way now to do it is to actually create the file while I am the tool, which then gives problems if I want to update the file.
- Thanks for the support! --beetstra (talk) 06:03, 29 January 2014 (UTC)
Please restart COIBot
Hi. COIBot is not listening to IRC commands or taking hours to reply to them. I wonder if you could restart the bot so it goes back to normal? This is happening since yesterday. FWIW I have access to tool labs and I'm a mantainer for Nova Resource:Tools.stewardbots. I wonder if I could get access to tools.coibot so when a restart is required I can do it myself if you explained me how do you handle this. Best regards, --MarcoAurelio (talk) 16:47, 24 November 2016 (UTC)