during sustained overload situation replies are truncated in production
Bug #636713 reported by
Robert Collins
This bug affects 2 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Launchpad itself |
Won't Fix
|
High
|
Unassigned |
Bug Description
See the 2010-09-
We seem to have a bug where:
- when LP is kept 100% busy (that is, 4 pending requests and 4 active ones, at all times)
- it is running behind HAproxy and apache
- clients will get requests (e.g. +activereviews, WADL, merge proposals) truncated.
One user (wgrant) reported the truncation as almost always being at around 12K of data.
Changed in launchpad-foundations: | |
status: | New → Triaged |
importance: | Undecided → High |
To post a comment you must log in.
Here are some example errors from one of my API scripts. Some for WADL, others for collections. The HTTP response headers for the 1.2MB WADL are about 350 bytes.
httplib. IncompleteRead: IncompleteRead( 15646 bytes read, 1272188 more expected) IncompleteRead: IncompleteRead( 15710 bytes read, 39088 more expected) IncompleteRead: IncompleteRead( 15710 bytes read, 35557 more expected) IncompleteRead: IncompleteRead( 15710 bytes read, 38408 more expected) IncompleteRead: IncompleteRead( 15646 bytes read, 1272188 more expected) IncompleteRead: IncompleteRead( 15646 bytes read, 1272188 more expected)
httplib.
httplib.
httplib.
httplib.
httplib.