stuff related to the mirroring infrastructure, protocol, and tooling used to maintain the Software Heritage mirror network
Details
Details
Description
May 11 2021
May 11 2021
vlorentz closed T3303: Replaying REMD topics is broken, a subtask of T3197: Mirror: fix common issues of a replayer session, as Resolved.
May 4 2021
May 4 2021
vlorentz added a revision to T3303: Replaying REMD topics is broken: D5670: Stop storing authority/fetcher metadata..
Apr 27 2021
Apr 27 2021
zack added a comment to T1576: document the typical cost(s) of hosting an archive mirror.
docs !
moranegg added a comment to T1576: document the typical cost(s) of hosting an archive mirror.
Where should this information be available?
- on the main website
- docs
- a private location
Apr 23 2021
Apr 23 2021
Apr 22 2021
Apr 22 2021
douardda added a comment to T3200: Mirror: year is out of range.
These revisions are probably coming from https://gitlab.com/gitlab-org/gitlab-test (or a clone)
douardda added a comment to T3200: Mirror: year is out of range.
Ah fun, one of the revisions with this pb, on staging (ba3343bc4fa403a8dfbfcab7fc1a8c29ee34bd69) seems to have been crafted by https://gitlab.com/gitlab-org/gitlab-foss/-/blob/staging-26-fix_add_deploy_key_spec/spec/models/merge_request_diff_commit_spec.rb
Apr 21 2021
Apr 21 2021
douardda added a comment to T3200: Mirror: year is out of range.
See T3170 (error generated by the same invalid kafka messages).
Apr 19 2021
Apr 19 2021
olasd closed T2003: Content replayer may try to copy objects before they are available from an objstorage, a subtask of T1914: Keep mirror of contents on S3 up to date, as Resolved.
Apr 8 2021
Apr 8 2021
douardda added a comment to T3198: Mirror: unexpected closed connection to the pg server.
Just got this one below. Note that this occurred just when the replayer actually started to insert object in the storage (before that, since the start of the replayer process, only kafka scaffolding took place for quite some time, around 30mn!)
Apr 6 2021
Apr 6 2021
douardda closed T3201: Mirror: unsupported Unicode escape sequence as Resolved by committing rDSTO39507b24d0f4: Make the replayer drop the Revision.metadata.
douardda closed T3201: Mirror: unsupported Unicode escape sequence, a subtask of T3197: Mirror: fix common issues of a replayer session, as Resolved.
Apr 2 2021
Apr 2 2021
douardda added a comment to T3197: Mirror: fix common issues of a replayer session.
Currently, the mirror test session is running with:
douardda added a comment to T3201: Mirror: unsupported Unicode escape sequence.
easy fix: modify the replayer to ignore this 'metadata' column while inserting revisions
douardda added a comment to T3201: Mirror: unsupported Unicode escape sequence.
09:45 <+vlorentz> douardda: yes and the only way around it (short of dropping data) is T3089 09:46 -swhbot:#swh-devel- T3089 (submitter: vlorentz, owner: vlorentz, status: Open): Remove the 'metadata' column of the 'revision' table <https://forge.softwareheritage.org/T3089> 09:46 <+vlorentz> or switching to cassandra 09:46 <+vlorentz> the good news is, they couldn't be inserted in the storage either, so you can safely drop them for now
douardda triaged T3199: Mirror: key value violates unique constraint "person_fullname_idx" as High priority.
Mar 15 2021
Mar 15 2021
Mar 11 2021
Mar 11 2021
rdicosmo updated the task description for T3116: Roll out first operational mirror.
rdicosmo added a parent task for T1576: document the typical cost(s) of hosting an archive mirror: T3116: Roll out first operational mirror.
rdicosmo added a subtask for T3116: Roll out first operational mirror: T3054: Scale out object storage design.
Mar 4 2021
Mar 4 2021
Dec 23 2020
Dec 23 2020
zack triaged T2914: mirror documentation: add ballpark storage/infra requirements as Normal priority.
Nov 17 2020
Nov 17 2020
The new cluster in rocquencourt is using the built-in Kafka ACLs now (9993a81ffc7a1c8bd519b33ae63ac1145105f624).
Oct 16 2020
Oct 16 2020
douardda added a comment to T2706: Benchmark objstorage for mirror (uffizi vs. azure vs. s3).
Same as before but with 1M (fresh) sha1s:
douardda added a comment to T2706: Benchmark objstorage for mirror (uffizi vs. azure vs. s3).
Since the results on uffizi above did suffer from a few caveats, I've made a few more tests:
- a first result has been obtained with a dataset that had only objects stored on the XFS part of the objstorage
- a second dataset has been created (with the order by sha256 part to spread the sha1s)
- but results are a mix hot/cold cache tests
Oct 15 2020
Oct 15 2020
douardda added a comment to T2706: Benchmark objstorage for mirror (uffizi vs. azure vs. s3).
Some results:
zack added projects to T2706: Benchmark objstorage for mirror (uffizi vs. azure vs. s3): Mirror, Object storage.
Sep 22 2020
Sep 22 2020
olasd added a comment to T1828: Improve directory journal backfill performance.
(the backfill had, in fact, completed within a month)
At this point, I don't think we'll make it much better with postgres as source.
moranegg moved T1576: document the typical cost(s) of hosting an archive mirror from Backlog to sponsors/clients on the Documentation board.
Apr 28 2020
Apr 28 2020
olasd closed T2350: Support large messages in swh.journal / kafka, a subtask of T2348: swh.journal silently loses large objects instead of rejecting them, as Resolved.
The kafka producer in swh.journal now reads message receipts and fails if they're negative, or if they didn't arrive within two minutes.
Apr 15 2020
Apr 15 2020
olasd changed the status of T2351: Consider backfilling mistakenly rejected large objects from PostgreSQL, a subtask of T2348: swh.journal silently loses large objects instead of rejecting them, from Open to Work in Progress.
olasd closed T2349: Make the journal writer reliable, a subtask of T2348: swh.journal silently loses large objects instead of rejecting them, as Resolved.
Apr 6 2020
Apr 6 2020
olasd triaged T2348: swh.journal silently loses large objects instead of rejecting them as High priority.