action #64809
closedWorker uploads some text results possibly multiple times wasting resources
0%
Description
Observation¶
Worker uploads some text results possibly multiple times.
I've created the following change track result uploads (and their size): https://github.com/os-autoinst/openQA/compare/master...Martchus:log-size-accounting?expand=1
With this modification one can see lots of duplicated uploads in the web UI log. I used the following command with the web UI log from a local test run:
sort webui-log-job-1260-opensuse-15.1-DVD-Updates-x86_64-Build20200321-1-gnome@uefi.log | uniq -d | grep -i accounting
(The log file is attached.)
Problem¶
This wastes resources mainly network bandwidth, increased disk I/O on the web UI host and time.
Expected result¶
Every result is only uploaded once unless a retry is necessary due to transfer problems.
Suggestions¶
- Ensure there is no duplicate upload, e.g. – if necessary – apply the same hash lookup algorithm as for screenshots
Notes / suggestion¶
For screenshots the web UI provides the worker with a list of known screenshot MD5 sums to avoid this. A similar mechanism could be implemented for text results.
Files
Updated by mkittler about 4 years ago
- Related to action #64574: Keep track of disk usage of results by job groups added
Updated by mkittler about 4 years ago
Updated by okurz about 4 years ago
- Description updated (diff)
- Category set to Regressions/Crashes
- Status changed from New to Workable
I updated the ticket to make it workable.
Updated by mkittler about 4 years ago
- Description updated (diff)
- Category deleted (
Regressions/Crashes) - Status changed from Workable to New
Reading your amended suggestions it seems you misunderstood the ticket. The provided log file is already from a local run. It does not need to be analyzed further to make conclusions. I only uploaded it here to make the observation I've already made more transparent.
I'd also like to note that we likely don't need to hash the text files. Of course it would be useful for a global de-duplication like we have for screenshots. The reason I opened this ticket is that currently the very same text result (the exact same step within the same job) is sometimes uploaded twice.
Updated by mkittler about 4 years ago
- Status changed from New to In Progress
- Assignee set to mkittler
Updated by mkittler about 4 years ago
- Status changed from In Progress to Resolved
The PR has been merged: https://github.com/os-autoinst/openQA/pull/2879