1148724 - scheduler retries too aggressively on error

Status: RESOLVED FIXED

(Firefox Graveyard :: Reading List, defect)

Description

[Mark Hammond [:markh] [:mhammond]]

Attachment: 0004-Bug-XXXXXXX-reading-list-scheduler-gets-a-back-off-s.patch

Currently the scheduler will retry a sync every 2 minutes until it sees success. This is too aggressive - if the server is overloaded or down this will cause every running profile to attack it :)  If the error is on the client side the user may notice increased network traffic and a performance hit. Later we should differentiate different types of error but until then we should be sensible.

This patch doubles the timeout on error with a max of the "normal" schedule - ie, first error will cause a 2 minute retry, then 4, 8, 16, 32 up to 7200 (default 2 hour schedule). Sequence restarts on next successful sync.  Note also we try and sync after any local RL item change regardless of the current schedule.

This is a fairly simple patch with tests. I think we should take it on 38 - in a worst-case scenario we could effectively DDOS our own service!

Comment 1

[Mark Hammond [:markh] [:mhammond]]

(In reply to Mark Hammond [:markh] from comment #0)
> This patch doubles the timeout on error with a max of the "normal" schedule
> - ie, first error will cause a 2 minute retry, then 4, 8, 16, 32 up to 7200
> (default 2 hour schedule).

Doh - mixed up my units - 7200 is seconds - max is 2 hours, 120 minutes.

Comment 2

[Mark Hammond [:markh] [:mhammond]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/0c64436fdc79
https://hg.mozilla.org/integration/fx-team/rev/b34ca540cf73

Comment 3

[Mark Hammond [:markh] [:mhammond]]

Landed on central as https://hg.mozilla.org/mozilla-central/rev/b34ca540cf73