action #174304
Updated by okurz about 2 months ago
Some LTP tests trigger large amounts of kernel log messages and then upload journalctl logfile to OpenQA. In this example, `shutdown_ltp-journalctl.txt` log file contains 18MB of data and the upload took 111 seconds (upload speed: ~168KB/s, upload timeout: 100s).
https://openqa.suse.de/tests/16073659#step/shutdown_ltp/55
The PowerVM SUT (upload source) is nessberry-2, the OpenQA controller host (upload destination) is worker29.oqa.prg2.suse.org. What is the expected bandwidth between the machines?
## Acceptance criteria
* **AC1:** A sufficiently high bandwidth is ensured between usual OSD involved testing machines, e.g. ~1GBit/s
## Suggestions
* Follow and respond on https://sd.suse.com/servicedesk/customer/portal/1/SD-175581
* Check bandwidth between various hosts, e.g. with iperf, and comment back in the SD ticket
* And use `-4` and `-6` to check IPv4 vs. IPv6
* Depending on how the problem can be reproduced, e.g. which hosts are affected, investigate/mitigate/solve the underlying issue, e.g. anything in the range from "just trigger a reboot to the HMC or switch" to work in a task force with IT to identify misbehaving systems severly impacting the available bandwidth
* Also be aware about #133700 and shove it in the face of IT
* Don't just bump timeouts here