Do not try executing operations on inactive PGs at all
This commit is contained in:
@@ -1317,7 +1317,8 @@ int cluster_client_t::try_send(cluster_op_t *op, int i)
|
||||
auto & pool_cfg = st_cli.pool_config.at(INODE_POOL(op->cur_inode));
|
||||
auto pg_it = pool_cfg.pg_config.find(part->pg_num);
|
||||
if (pg_it != pool_cfg.pg_config.end() &&
|
||||
!pg_it->second.pause && pg_it->second.cur_primary)
|
||||
!pg_it->second.pause && pg_it->second.cur_primary &&
|
||||
(pg_it->second.cur_state & PG_ACTIVE))
|
||||
{
|
||||
osd_num_t primary_osd = pg_it->second.cur_primary;
|
||||
if (pool_cfg.local_reads != POOL_LOCAL_READ_PRIMARY &&
|
||||
|
||||
Reference in New Issue
Block a user