Details
-
Bug
-
Resolution: Fixed
-
Critical
-
2.0.1
-
Security Level: Public
-
centos 6.4 x86_64
-
Triaged
-
PCI Team - Sprint 8
Description
I'm trying to reach 32000 requests per second and above. It is possible with memcached but impossible with moxi+couchbase. Segfaults are constantly appearing in moxi on high load for some reason. I have core dumps but moxi doesn't have any debug info. I used moxi binary from couchbase community edition 2.0.1
The moxi configuration I used:
port_listen = 11220,
default_bucket_name = default,
connect_timeout = 400,
connect_max_errors = 5,
wait_queue_timeout = 200,
downstream_timeout = 5000,
downstream_max = 1024,
downstream_conn_max = 4,
downstream_conn_queue_timeout = 200,
#connect_rerty_interval = 300,
cycle = 200
/opt/couchbase/bin/moxi -Z /opt/moxi/etc/moxi.cfg -z /opt/moxi/etc/moxi-cluster.cfg -c 12288 -t 12
Attachments
Issue Links
- relates to
-
CCBC-724 >1024 selectors
- Closed
For Gerrit Dashboard: MB-8102 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
30345,6 | MB-8102: Switched from using select(..) to poll(..) within mcs_io_read | master | moxi | Status: MERGED | +2 | +1 |
30424,1 | MB-8102: fix moxi segfault due to overflowed file descriptors | master | manifest | Status: ABANDONED | +2 | +1 |
30431,2 | MB-8102: fix moxi segfault due to overflowed file descriptors | master | manifest | Status: MERGED | +2 | +1 |
30433,2 | MB-8102: Switched from using select(..) to poll(..) within mcs_io_read | 2.5.0 | moxi | Status: MERGED | +2 | +1 |
30434,2 | MB-8102: use branch to pull in just the changes needed for segfault hot-fix | master | manifest | Status: MERGED | +2 | +1 |
30798,2 | MB-8102: use commits from released/1.8.1.xml | master | manifest | Status: MERGED | +2 | +1 |
32217,1 | MB-8102: resurrect 1.8.1 builders | master | build | Status: MERGED | 0 | 0 |
32218,1 | MB-8102: fix 181 ubuntu-64 builder name | master | build | Status: MERGED | 0 | 0 |