configured retry-delay for MPI has no (or very minimal) effect

[pautri]

the currently configured retry-delay for the MPI provider does not seem to have an effect. Retry timestamps of the last run (before switching to the OLAC fallback after 5 failed attempts):

2019-07-25T18:33:01,080
2019-07-25T18:33:01,488
2019-07-25T18:33:01,892
2019-07-25T18:33:02,370
2019-07-25T18:33:03,076

If the configured values are in milliseconds, they are a rather small change compared to the average interval of 400-500 ms between requests.

[twagoo]

Had a closer look; it seems that the right configuration makes it to the production instance and is partially applied: while default retry count is 2, the MPI provider gets 5 retries. However the retry-delay settings seems to be ignored.

Config line:

oai-harvest-config/config-clarin-clarin.xml

Line 112 in 207a1a5

<config url="https://archive.mpi.nl/oai2" max-retry-count="5" retry-delay="10 30 60 360"/>

[menzowindhouwer]

Retry delays where in milliseconds, so 10 30 60 360 was faily short. Changed now to seconds inline with the timeout configuration.

3f908aa

menzowindhouwer/oai-harvest-manager@9582300