Hate to necro an old thread. This issue went away for the most part for about 6-8 months but now is back with a vengeance. Its happening almost daily now.
On Sun, Dec 4, 2016 at 2:08 PM, John phoenixoverride@gmail.com wrote:
Via the Python log rotation process. I've been using the same code for years though.
On Sun, Dec 4, 2016 at 1:14 PM Tim Landscheidt tim@tim-landscheidt.de wrote:
(anonymous) wrote:
Correct, when I notice files with 0000 I fix them. However it's
happening
repeatedly and just started recently.
[…]
How are those files created?
Tim
Labs-l mailing list Labs-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/labs-l
On Sun, Oct 15, 2017 at 8:16 PM, John phoenixoverride@gmail.com wrote:
Hate to necro an old thread. This issue went away for the most part for about 6-8 months but now is back with a vengeance. Its happening almost daily now.
I'm not aware of any systemic changes to our NFS server setup on the client or server side in say the past 30 days. It seems like we would have reports from more than your tool if new file creation was broadly broken. That does not mean that your problem is unimportant, but it makes it more likely that it is somehow related to code in use by your tool.
This sounds like an issue that should be written up as a Phabricator task with relevant information from the past emails and current problem summarized and if possible some minimum code to reproduce.
Bryan
How is the python script that creates the files being run? What is its umask?
I suspect it may run under different conditions, and sometimes end up with a 777 umask.