slow mirrored branches can block branch puller
Bug #520107 reported by
Michael Hudson-Doyle
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Launchpad itself |
Fix Released
|
High
|
Michael Hudson-Doyle | ||
Bug Description
We've had a few instances of network glitches causing mirrored branches on a particular host to time out, and if the puller ends up trying to pull lots of branches from the one host at once, the puller clogs up.
One "fix" would be to have a separate puller process for mirrored branches again, which would let mirrored branches rot in their own corner and let hosted branches be processed normally.
More outlandish schemes involve scaling timeouts in various ways to fail branches more quickly most of the time.
Related branches
lp:~mwhudson/launchpad/pull-mirror-branches-separately-bug-520107
- Tim Penhey (community): Approve
-
Diff: 527 lines (+144/-46)14 files modifiedcronscripts/supermirror-pull.py (+3/-1)
lib/lp/code/errors.py (+5/-0)
lib/lp/code/interfaces/branch.py (+0/-4)
lib/lp/code/interfaces/branchpuller.py (+3/-1)
lib/lp/code/interfaces/codehosting.py (+4/-1)
lib/lp/code/model/branchpuller.py (+5/-2)
lib/lp/code/model/tests/test_branchpuller.py (+47/-9)
lib/lp/code/xmlrpc/codehosting.py (+11/-2)
lib/lp/code/xmlrpc/tests/test_codehosting.py (+17/-9)
lib/lp/codehosting/inmemory.py (+12/-2)
lib/lp/codehosting/puller/scheduler.py (+3/-2)
lib/lp/codehosting/puller/tests/test_acceptance.py (+19/-2)
lib/lp/codehosting/puller/tests/test_scheduler.py (+14/-10)
lib/lp/testing/factory.py (+1/-1)
| Changed in launchpad-code: | |
| assignee: | nobody → Michael Hudson (mwhudson) |
| Changed in launchpad-code: | |
| status: | Triaged → In Progress |
| Changed in launchpad-code: | |
| status: | In Progress → Triaged |
| status: | Triaged → Fix Committed |
| Changed in launchpad-code: | |
| status: | Fix Committed → Fix Released |
To post a comment you must log in.
