Start old cassandra and scylla nodes one by one #657
Conversation
fruch
left a comment
fruch
left a comment
There was a problem hiding this comment.
this logic should be in whom use CCM
those parameters are not really related to parallel execution, they are they for testing purposes to error paths.
I don't think this logic should be in CCM
Added |
dkropachev
force-pushed
the
dk/fix-cassandra-cluster-start
branch
from
59c7cae to
7127f24
Compare
Cassandra 3.0 and Scylla older than 6.x does not support parallel node initialization. Tune CCM to start nodes one by one for it.
dkropachev
force-pushed
the
dk/fix-cassandra-cluster-start
branch
from
7127f24 to
b2b9ac9
Compare
|
|
||
| @property | ||
| def parallel_start_supported(self): | ||
| return self.database_version.startswith('3.') |
| self.default_wait_for_binary_proto = 420 if self.scylla_mode != 'debug' else 900 | ||
|
|
||
| @property | ||
| def parallel_start_supported(self): |
There was a problem hiding this comment.
the logic about version support, should be in the test/framework trying to enable it.
the versions use here are not even correct, it was supported much before 2025.1.
again the support matrix for thing kind of thing, should be in here.
| if jvm_args is None: | ||
| jvm_args = [] | ||
| if parallel_start is None: | ||
| parallel_start = self.parallel_start_supported |
There was a problem hiding this comment.
this should be explicit, not like that.
this gonna break all of dtest
There was a problem hiding this comment.
why ? we already start scylla nodes in parallel
|
@dkropachev what is the context of this one, why are you trying to change the default behavior, for what end exactly ? just cause you don't like the |
Context is that I have tried to start cassandra cluster with many nodes and couple of times it failed with token ring collission, due the fact that we are starting them in parallel. |
So pass the wait_for parameters and you are fine. Why do you need automatically do that decision? |
2 participants
Cassandra 3.0 and Scylla older than 6.x does not support parallel node initialization.
Tune CCM to start nodes one by one when server does not support parallel node initialization.