Patchbot r640 fails with "disk full"

36 views
Skip to first unread message

Samuel Lelievre

unread,
Jul 27, 2020, 10:49:06 PM7/27/20
to sage-devel
Patchbot r640 seems to be failing for lack of disk space:



Markus Wageringel

unread,
Jul 28, 2020, 4:26:45 AM7/28/20
to sage-devel
Thank you for the note. I have disabled this patchbot client for now.

The problem seems to be that the /tmp directory gets flooded with thousands of little files that do not get removed automatically. It has been less than two weeks since I have last cleared the /tmp directory, and now there are already more than 100.000 files. Is there a way to let the patchbot delete the temporary files it created automatically after each ticket? The machine the patchbot client is running on has recently been switched from CentOS to Ubuntu – maybe it is just a coincidence, but this has not been a problem before.

97% of the temporary files are created by ECL (they are named `ecl....`). Maybe the ECL upgrade in 9.2.beta5 has something to do with it?

Another major problem are two huge files containing the following output. One of them is more than 30 GB in size. Does anyone know where this is coming from? This code is not the source tree of the develop branch, so it might be some trac ticket, possibly related to Cysignals.

sage: cython('while True: print("Hello")') ## line 245 ##
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
Hello
...

Samuel Lelievre

unread,
Jul 28, 2020, 7:04:08 AM7/28/20
to sage-devel
Possibly related:


Feel free to open another issue or to send a pull request
if you see how to fix any known patchbot issue.

Samuel Lelievre

unread,
Jul 28, 2020, 7:09:25 AM7/28/20
to sage-devel
If your patchbot runs only at certain hours of the day,
you could also clear /tmp daily before patchbot starts,
using a cronjob.

Markus Wageringel

unread,
Jul 28, 2020, 8:29:16 AM7/28/20
to sage-...@googlegroups.com
Yes, I will give that a try. Thanks.

By searching through the patchbot logs in reverse, I also found the ticket that was causing the infinite print loop: https://trac.sagemath.org/ticket/19025.

Reply all
Reply to author
Forward
0 new messages