Suggestion - indicate failed tasks
Moderators: Site Moderators, FAHC Science Team
-
- Posts: 308
- Joined: Wed Feb 16, 2022 1:18 am
Suggestion - indicate failed tasks
When I used HFM, it told me how many failed tasks and how many successfull tasks for each slot. This is usefull as I can see at a glance if a GPU is causing errors, being useless, wasting the scientists's time and needs tending to. Could this be added to the Beta? Some kind of notification something has gone wrong. "Previous task failed" next to the resource would alert me to go check the log etc. Better if the notification then disappeared with a successfull task.
Re: Suggestion - indicate failed tasks
I mentioned something similar here in GitHub: Issue #53 and added a link to your post.
Currently, with the web interface you can view the log file, and filter to see Errors or Warnings (check boxes in the upper-right of the log viewer)
Currently, with the web interface you can view the log file, and filter to see Errors or Warnings (check boxes in the upper-right of the log viewer)
-
- Posts: 308
- Joined: Wed Feb 16, 2022 1:18 am
Re: Suggestion - indicate failed tasks
Thanks. I only have some minor warnings.
Most of my machines are saying:
01:32:34:E :Exception: Failed to open dynamic library 'nvcuda.dll': The specified module could not be found.
But I don't have any cuda cards anywhere.
They're also saying:
00:00:46:E :Exception: In thread 3: Failed to rename 'log.txt' to 'logs\log-20230210-000046.txt': The process cannot access the file because it is being used by another process.
Most of my machines are saying:
01:32:34:E :Exception: Failed to open dynamic library 'nvcuda.dll': The specified module could not be found.
But I don't have any cuda cards anywhere.
They're also saying:
00:00:46:E :Exception: In thread 3: Failed to rename 'log.txt' to 'logs\log-20230210-000046.txt': The process cannot access the file because it is being used by another process.
-
- Site Moderator
- Posts: 1115
- Joined: Sat Dec 08, 2007 1:33 am
- Location: San Francisco, CA
- Contact:
Re: Suggestion - indicate failed tasks
The exception loading cuda is harmless and expected if you don’t have nvidia software installed.
The log error looks like you either have two clients running, or a client was killed abruptly and a file lock wasn’t released.
The log error looks like you either have two clients running, or a client was killed abruptly and a file lock wasn’t released.
-
- Posts: 308
- Joined: Wed Feb 16, 2022 1:18 am
Re: Suggestion - indicate failed tasks
I can't see any reason two clients would have been running. They certainly aren't now.
Unfortunately the log doesn't show the date, only the time. Not very helpful! I know, if you're viewing the whole log, you can tediously scroll up to find the last date stamp, but that doesn't work when you've filtered.
Unfortunately the log doesn't show the date, only the time. Not very helpful! I know, if you're viewing the whole log, you can tediously scroll up to find the last date stamp, but that doesn't work when you've filtered.
-
- Site Moderator
- Posts: 1115
- Joined: Sat Dec 08, 2007 1:33 am
- Location: San Francisco, CA
- Contact:
Re: Suggestion - indicate failed tasks
If you’re running 8.1.13 the default log rotation is 24 hours.
-
- Posts: 308
- Joined: Wed Feb 16, 2022 1:18 am
Re: Suggestion - indicate failed tasks
In that case one machine had the log access complaint just after midnight. I can think of no reason anything was going on then. Could it be if the log is being read by Chrome to display it on another machine? It's a pity it doesn't say what the other process is!
-
- Site Moderator
- Posts: 1115
- Joined: Sat Dec 08, 2007 1:33 am
- Location: San Francisco, CA
- Contact:
Re: Suggestion - indicate failed tasks
I don’t know. I’m not a Windows expert.