`Iceberg` + `object_storage_cluster`: SELECT returns empty result (after ~60s) when all ClickHouse nodes are SIGSTOP’ed instead of throwing an error

[alsugiliazova]

Describe the bug
When querying an Iceberg table via a DataLakeCatalog database and specifying SETTINGS object_storage_cluster, if ClickHouse is paused (SIGSTOP) on all swarm nodes, the query completes after ~60 seconds with 0 rows instead of raising an exception that no nodes are available.

To Reproduce
I used 25.3.3.20186.altinityantalya.

1. I used DataLakeCatalog database engine
2. Pause ClickHouse on all Swarm nodes:

pgrep -x clickhouse | xargs -r kill -STOP

Unpause ClickHouse after running select query:

pgrep -x clickhouse | xargs -r kill -CONT

After unfreeze everything works as expected.
3. Query:

clickhouse1 :) SELECT
    count(),
    hostName()
FROM datalakecatalog_db_948b9fb6_829c_11f0_953b_e0c26496f172.`namespace_948ba911_829c_11f0_910d_e0c26496f172.table_948ba93a_829c_11f0_b866_e0c26496f172` 
GROUP BY hostName()
SETTINGS object_storage_cluster = 'static_swarm_cluster'

SELECT
    count(),
    hostName()
FROM datalakecatalog_db_948b9fb6_829c_11f0_953b_e0c26496f172.`namespace_948ba911_829c_11f0_910d_e0c26496f172.table_948ba93a_829c_11f0_b866_e0c26496f172`
GROUP BY hostName()
SETTINGS object_storage_cluster = 'static_swarm_cluster'

Query id: eef53ad1-696b-4df9-848e-b9ed88f2d5ed

Ok.

0 rows in set. Elapsed: 60.030 sec. 

Expected behavior
An exception indicating that none of the nodes are available.

[ianton-ru]

The same in upstream with s3Cluster function, fix for upstream - ClickHouse#86297

[ianton-ru]

Upstream merged alternative PR - ClickHouse#86414
TODO: backport it into Antalya instead of #982