TokuMX 2.0.0 Crash

Hi,

We have a cluster where the master runs TokuMX 2.0.0, and one of the slaves runs TokuMX 2.0.1 (probably not a good practice…)

I connected to the shell of the slave, and it seems the master crashed just around this time:

“5f64fd2a9871e47987d4541c8840e60a” }
Wed Feb 17 07:46:57.476 [conn2025425] warning: No such role, “userAdminAnyDatabase”, in database placer-production. No privileges will be acquired from this role
Wed Feb 17 07:46:58.198 [conn1904435] authenticate db: admin { authenticate: 1, nonce: “79e99c26ec13f30a”, user: “", key: “a7397abae87761b31f6e50507a11a44c” }
Wed Feb 17 07:46:58.198 [conn1750423] authenticate db: admin { authenticate: 1, nonce: “6e71536c7df5c038”, user: "
”, key: “7170a9969616979645b03eeb494f4852” }
Wed Feb 17 07:46:58.362 [conn2025414] end connection 104.197.100.112:50743 (195 connections now open)
Wed Feb 17 07:46:58.364 [initandlisten] connection accepted from 104.197.100.112:50748 #2025426 (196 connections now open)
/data/package-deb-debian-wheezy/build/tokumx-2.0.0/src/third_party/ft-index/ft/bndata.cc:362 get_memory_size: Assertion `retval >= get_disk_size()’ failed (errno=0)
: Success

Not sure if this is related to the different versions, but anyway looks like something you’d want to investigate.

Thanks,
Zohar Bar-Yehuda

Hi Zoharby,

Good practices point that master / slaves should be in the same version of TokuMX.

Could you also upload logs from master and secondary ?

Thanks