- **status**: review --> fixed
---
** [tickets:#1579] imm: sImplDetachTime is not cleared when aborting sync**
**Status:** fixed
**Milestone:** 4.5.2
**Created:** Fri Oct 30, 2015 10:28 AM UTC by Hung Nguyen
**Last Updated:** Fri Oct 30, 2015 01:18 PM UTC
**Owner:** Hung Nguyen
SC-2 was the coordinator.
After the sync started:
Aug 20 11:32:01 PL-2-10 osafimmnd[6028]: NO NODE STATE->
IMM_NODE_W_AVAILABLE
Aug 20 11:32:01 PL-2-10 osafimmnd[6028]: NO SERVER STATE:
IMM_SERVER_SYNC_PENDING --> IMM_SERVER_SYNC_CLIENT
...some implementers from SC-2 were set:
Aug 20 11:32:07 PL-2-10 osafimmnd[6028]: NO Implementer (applier)
connected: 17 (@safPmService1315992) <0, 2020f>
Aug 20 11:32:07 PL-2-10 osafimmnd[6028]: NO Implementer (applier)
connected: 18 (@safPmService1315991) <0, 2020f>
Then SC-2 went down, and there was a global discard node message
(immnd_evt_proc_discard_node):
Aug 20 11:32:22 PL-2-10 osafimmnd[6028]: NO Global discard node received
for nodeId:2020f pid:8153
In ImmModel::discardNode method, implementers from SC-2 were set to
sImplDetachTime (6 seconds).
IMPORTANT: sImplDetachTime has first field type (ImplementerInfo *)
Since PL-10 was in SYNC state, the sync is aborted due to the disappearance of
the coordinator (immnd_evt_proc_abort_sync)
Aug 20 11:32:22 PL-2-10 osafimmnd[6028]: WA Global ABORT SYNC received for
epoch 143
Aug 20 11:32:22 PL-2-10 osafimmnd[6028]: WA SERVER STATE:
IMM_SERVER_SYNC_CLIENT --> IMM_SERVER_LOADING_PENDING (sync aborted)
In ImmModel::abortSync(), all implementers (ImplementerInfo struct) are deleted:
if(!sImplementerVector.empty()) {
ImplementerVector::iterator i;
for(i = sImplementerVector.begin(); i !=
sImplementerVector.end();
++i) {
ImplementerInfo* info = (*i);
delete info;
}
sImplementerVector.clear();
}
The upper code also deleted attached implementers from SC-2, but implemeters
from SC-2 still remain in sImplDetachTime map.
That’s why IMMND crashed at line 12643 when it tried to access the field of
deleted ImplementerInfo struct:
/* Remove old implementer detach times. */
iem=sImplDetachTime.begin();
while(iem!=sImplDetachTime.end()) {
if(now - iem->second.mCreateTime >= DEFAULT_TIMEOUT_SEC) {
TRACE_5("Timeout on sImplDetachTime implid:%u", iem->first->mId);
sImplDetachTime.erase(iem);
iem=sImplDetachTime.begin();
} else {
++iem;
}
}
---
Sent from sourceforge.net because [email protected] is
subscribed to https://sourceforge.net/p/opensaf/tickets/
To unsubscribe from further messages, a project admin can change settings at
https://sourceforge.net/p/opensaf/admin/tickets/options. Or, if this is a
mailing list, you can unsubscribe from the mailing list.------------------------------------------------------------------------------
_______________________________________________
Opensaf-tickets mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets
