Closed kbr-scylla closed 1 year ago
@nyh - could this be due to https://github.com/scylladb/scylladb/commit/0e602159b9324997ca876f61b98a9276cebe0e1e ?
https://github.com/scylladb/scylladb/commit/0e602159b9324997ca876f61b98a9276cebe0e1e could only improve startup times.
The log message in CCM itself was introduced recently. commit 0184a10183567dac1bb5b8b16b2bb941f22fb560
Author: Israel Fruchter <fruch@scylladb.com>
Date: Thu May 25 12:12:01 2023 +0300
ccmlib/scylla_node: wait for scylla process to be running
seem like there was a call to `node.is_running()` that
wasn't checking the output, and assuming it would update the pid
information, and wasn't always the case.
this was causing cases that the next node would identifed
as running and not considered as part of the `wait_other_notice`
check.
we just need to polish it a bit (add rate-limiting etc.)
@nyh - could this be due to scylladb/scylladb@0e60215 ?
I don't understand how - look at that code, it doesn't print anything between each iteration.
It seems more reasonable that we have some recently-changed code in dtest or ccm that prints this silly message. Try grepping this string in ccm or dtest.
The log message in CCM itself was introduced recently. commit 0184a10
Right, this makes more sense.
we just need to polish it a bit (add rate-limiting etc.)
Or just print the message once (or once when starting to wait plus when the wait is done). Printing such a message every 0.1 seconds or every 1 seconds is both silly.
wait_for
, shouldn't be printing that excessively, or printing to stdout at all
Huge spam in the logs when I run dtest locally:
is it really necessary to print this every 10 milliseconds? I think this should be rate limited to ~once every 5 seconds or even less frequently, and maybe not printed in the main console log.