Table with 'compact storage' is not shown in "describe table" output in cqlsh

2022-09-27 Thread manish khandelwal
Hi All

As I understand there was a plan to drop *Compact Storage* support
with *Cassandra
4* but later few issues were identified which resulted in continued support
for Compact Storage in Cassandra 4. My cluster with a few old "compact
storage" tables  was able to come up with Cassandra 4.0.5. But when I
described those tables it gave me a warning



*Warning: Table keyspace_name.table_name omitted because it has constructs
not compatible with CQL (was created via legacy API).Approximate structure,
for reference:(this should not be used to reproduce this schema*)

The original schema is not printed. I find no issue querying the table,
only the schema is not shown.* Is it the expected behaviour for Cassandra 4
for compact tables? *

 DROP COMPACT STORAGE is not stable and marked experimental , another
option is to migrate tables but that might require changes at client side
as well.

Regards
Manish


Log messages "MUTATION messages were dropped" in system.log

2022-09-27 Thread Michail Kotsiouros via user
Hello Cassandra community,
I see the following log lines in the Cassandra nodes of 2 datacenters that 
participate in a cluster of 6 datacenters.

INFO  [ScheduledTasks:1] 2022-09-12 17:19:44,422 MessagingService.java:1236 - 
MUTATION messages were dropped in last 5000 ms: 0 internal and 2 cross node. 
Mean internal dropped latency: 0 ms and Mean cross-node dropped latency: 9140 ms

>From the information I have found online, I see that those logs indicate a 
>resource issue on the Cassandra nodes. I am trying to troubleshoot what could 
>be the bottleneck.
Can you, please, help me interpret the log line.
So far I know that
MUTATION means that it is write operations for already existing keys.
What internal means? Does it indicate that the write operation is originated 
from a coordinator node residing in the same datacenter?
What cross node means? Does it indicate that the write operation is originated 
from a coordinator node residing in another datacenter?
What dropped latency means? Is it the time that the message spent in some queue 
before dropped?

I would like to know that so as to understand the impact observed on client 
application.

Thank you in advance!!! I find really useful all the information found in 
different posts!!!

BR
Michail