OPERATIONAL DEFECT DATABASE
...
...
BugZero found this defect 2744 days ago.
I'm trying to add an index on a largish (~100 GB) and busy collection (many servers connecting, hundreds requests per second): db.screenshots.createIndex( Unknown macro: {"user_id"} , Unknown macro: {"unique"} , Unknown macro: {background} ); db.screenshots.createIndex( Unknown macro: {"deleted"} , Unknown macro: {"unique"} , Unknown macro: {background} ); It seems to work, but unfortunately after the index is built, all mongos instances on all servers crash (to be precise - around 90% of mongos processes crashed; the ones which didn't crash were on idle servers with few connections). We are using mongo 3.2.13 on Ubuntu 16.04, with sharding and replication, and mongos connecting. See the attached log with a crash.
kaloian.manassiev commented on Thu, 8 Jun 2017 06:37:45 +0000: Hi tch@virtall.com, Thank you for reporting this issue. This is a known problem, which was introduced in version 3.2.13 and has already been fixed under SERVER-29152. The fix will be available in version 3.2.14. If this is causing you recurrent problems, please downgrade to 3.2.12 until 3.2.14 is released, which should be very soon. Once again, thank you for your report and apologies for the inconvenience. Best regards, -Kal. tch@virtall.com commented on Thu, 8 Jun 2017 06:28:47 +0000: It seems that Jira garbled createIndex - let me know if I should readd them (don't see a way to edit the description).
