Analysis :
Refering confguration given in the description.
During lock operation of SU4 on PL-4, its component delays remove callback. But
by this time SU5 on PL-5 has already become active. Now PL-5 is rebooted, AMFD
performs node-failover of PL-5 and as a part of failover of active SU5 send
deletion of assignment to SU4 on PL-4. Since AMFND on PL-4 is already
processing removal of SU4, this deletion of assignment gets buffered. Actually
AMFD should not send redundant deletion of assignments. When comp of SU4
responds for the removal of assignments, AMFND deletes its SUSIs. After this
AMFND picks buffered deletion of SU and sets relevant flag for the processing
of assignments and calls the removal of assignment logic. Since SUSIs are
removed no callbacks are invoked and hence no response is sent to AMFD.
Generally, while sending response to AMFD all assignment related flags are
cleared by AMFND. But in this case they remained set. Now when SU4 was unlocked
for fresh assignments. After completion of assignments, avnd_su_si_oper_done()
logic
is called for all SI (SI pointer null) instead of single SI because of
uncleared flags due to buffered assignmets. While sending response to AMFD for
this fresh assignment, AMFND asserts as it does not find SI. Fresh assignments
are given one SI at a time and not for all the SIs simultaneouly like
modification. So AMFND expects SI pointer a vaild one.
Root cause of the problem is AMFD is sending redundant deletion of assignment
to AMFND. It must be fixed.
---
** [tickets:#1772] AMF: amfnd false assert when unlock 2N Active SU**
**Status:** accepted
**Milestone:** 4.7.2
**Created:** Wed Apr 20, 2016 02:39 AM UTC by Minh Hon Chau
**Last Updated:** Fri May 06, 2016 06:47 AM UTC
**Owner:** Praveen
**Attachments:**
-
[amfnd_assert.tgz](https://sourceforge.net/p/opensaf/tickets/1772/attachment/amfnd_assert.tgz)
(290.1 kB; application/x-compressed)
- Configuration: as attached file
. Set up 2N amf_demo with active SU4 on PL4, standby SU5, SU5B on PL5.
. 3 components for each SU
. 3 SIs, with dependency: SI3 -> SI2 -> S1 (highest sponsored SI)
- Steps:
. Bring up 2 SU
. Lock Active SU4, delay csi remove callback
. Meanwhile, reboot PL5
. Release delay of csi remove callback in SU4 -> assignments are removed
successfully
. PL5 restarts, SU5B get Active, SU5 get Standb assignments
. Lock SU5, SU5B
. Unlock SU4
-> amfnd on PL4 asserts
2016-04-20 12:36:14 PL-4 osafamfnd[423]: di.cc:835: avnd_di_susi_resp_send:
Assertion 'si' failed.
2016-04-20 12:36:14 PL-4 osafamfwd[465]: Rebooting OpenSAF NodeId = 0 EE Name =
No EE Mapped, Reason: AMF unexpectedly crashed, OwnNodeId = 132111,
SupervisionTime = 60
2016-04-20 12:36:14 PL-4 amf_demo[660]: AL AMF Node Director is down, terminate
this process
---
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.
------------------------------------------------------------------------------
Mobile security can be enabling, not merely restricting. Employees who
bring their own devices (BYOD) to work are irked by the imposition of MDM
restrictions. Mobile Device Manager Plus allows you to control only the
apps on BYO-devices by containerizing them, leaving personal data untouched!
https://ad.doubleclick.net/ddm/clk/304595813;131938128;j
_______________________________________________
Opensaf-tickets mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets
