This test has been running for a little under the timeout limit for the slower architectures, and at least for hwe-5.8 the changes introduced after -36 could not explain any change on the execution time of this testcase.
Therefore, I believe these failures have been caused by variations on the test systems and the testcase timeout should be adapted for expected longer runs.
This testcase hasn't been changed in several months, so we can discard any recent testcase change causing the timeout.
However, looking at the logs for focal/linux-hwe-5.8 runs on arm64 there are some interesting data.
The last time the testcase was successful was with version 5.8.0-36. 40~20.04. 1: /objectstorage. prodstack4- 5.canonical. com/v1/ AUTH_77e2ada1e7 a84929a74ba3b87 153c0ac/ autopkgtest- focal/focal/ arm64/l/ linux-hwe- 5.8/20210108_ 133824_ 09617@/ log.gz
https:/
Looking at the timestamps the test took a little bit over 19 min to finish.
Starting from version 5.8.0-37.42~20.04.1 it stated to fail: /objectstorage. prodstack4- 5.canonical. com/v1/ AUTH_77e2ada1e7 a84929a74ba3b87 153c0ac/ autopkgtest- focal/focal/ arm64/l/ linux-hwe- 5.8/20210113_ 031753_ cd06b@/ log.gz
https:/
The timeout reached was 1200 sec, i.e. 20 min.
This test has been running for a little under the timeout limit for the slower architectures, and at least for hwe-5.8 the changes introduced after -36 could not explain any change on the execution time of this testcase.
Therefore, I believe these failures have been caused by variations on the test systems and the testcase timeout should be adapted for expected longer runs.