lp:~hloeung/ubuntu-repository-cache/dont-disable-unit-on-failover-or-pause

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:ubuntu-repository-cache at revision 326

Paul Collins: Approve (lgtm) on 2021-03-07

Canonical IS Reviewers: Pending requested 2021-03-07

Diff: 99 lines (+16/-24)

2 files modified

lib/ubuntu_repository_cache/apache.py (+0/-5)
tests/unit/test_apache.py (+16/-19)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

326. By Haw Loeung on 2021-03-07

Don't set health-check-disabled flag in apache.create_metadata_site()

On non-leader units, metadata sync calls service.pause() which then
calls this. This causes a brief period where the unit is disabled and
not serving requests. It's useful for initial unit provisioning but
not so on a fully functional unit where metadata points to an existing
copy on disk.

325. By Haw Loeung on 2021-03-03

[trivial] Minor tidy up

324. By Haw Loeung on 2021-03-03

Log overrall check result and URLs for errors

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/health-check-logging-improvements/+merge/399130
Reviewed-by: Colin Misare <email address hidden>

323. By Haw Loeung on 2021-03-03

Avoid holding up connection/request based on sending metrics and flush buffers - LP:1917317

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/health-check-flush-output-buffers/+merge/399129
Reviewed-by: David Lawson <email address hidden>

322. By Haw Loeung on 2021-03-03

Send metrics at the very end - LP:1917317

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/health-check-send-metrics-at-end/+merge/399128
Reviewed-by: Alexandre Gomes <email address hidden>

321. By Haw Loeung on 2021-03-03

Include syncing dist-upgrade-all metadata - LP:1765934

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/sync-dist-upgrade-all-metadata/+merge/398938
Reviewed-by: Colin Misare <email address hidden>

320. By Haw Loeung on 2021-03-02

We're already logging, so may as well log to the right place - LP:1917385

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/fix-squid-access-log-location/+merge/399035
Reviewed-by: Paul Collins <email address hidden>

319. By Haw Loeung on 2021-03-02

Add socket timeout for sending metrics to influx

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/add-send-to-influx-socket-timeout/+merge/398931
Reviewed-by: Paul Collins <email address hidden>

318. By Haw Loeung on 2021-03-01

Add apache2 cgi script timeout

Otherwise, value of `Timeout` is used and that defaults to 60secs which is too long. We have two check points (metadata and pool) with a timeout in the health-check cgi script to 10secs for each so 20secs should be enough for both.

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/apache2-cgi-script-timeout/+merge/398920
Reviewed-by: Benjamin Allot <email address hidden>

317. By Haw Loeung on 2021-03-01

Bump metadata sync timeout

As seem in AWS us-east-1, it can take over an hour to sync the
metadata (which is now 26GiB) from another unit in the same region.

Reviewed-on: https://code.launchpad.net/~hloeung/ubuntu-repository-cache/bump-metadata-sync-timeout/+merge/398848
Reviewed-by: Max Horlanchuk <email address hidden>