[tickets] [opensaf:tickets] #2557 dtm: Use edge-triggered epoll

[Anders Widell via Opensaf-tickets]

---

** [tickets:#2557] dtm: Use edge-triggered epoll**

**Status:** unassigned
**Milestone:** 5.17.10
**Created:** Wed Aug 16, 2017 12:26 PM UTC by Anders Widell
**Last Updated:** Wed Aug 16, 2017 12:26 PM UTC
**Owner:** nobody


To improve performance, DTM should use edge-triggered epoll. Some mechanism to 
guarantee fairness is probably needed for this to work well, and should be 
implemented at the same time.


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2556 split-brain should be reported by a ER or CR in syslog

[Per Rodenvall via Opensaf-tickets]

---

** [tickets:#2556] split-brain should be reported by a ER or CR in syslog**

**Status:** unassigned
**Milestone:** 5.17.10
**Created:** Wed Aug 16, 2017 11:57 AM UTC by Per Rodenvall
**Last Updated:** Wed Aug 16, 2017 11:57 AM UTC
**Owner:** nobody


This was seen in syslog:
2017-06-27 13:49:32 SC-1 osafrded[176]: Rebooting OpenSAF NodeId = 0 EE Name = 
No EE Mapped, Reason: Split-brain detected, OwnNodeId = 131343, SupervisionTime 
= 60

In osaftest we look for ER and CR after each test. We in QuAT team really like 
split-brain to be a ER or CR


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2555 smf: execLevel for balanced upgrade

[Rafael Odzakow via Opensaf-tickets]

---

** [tickets:#2555] smf: execLevel for balanced upgrade**

**Status:** unassigned
**Milestone:** 5.17.10
**Created:** Wed Aug 16, 2017 11:39 AM UTC by Rafael Odzakow
**Last Updated:** Wed Aug 16, 2017 11:39 AM UTC
**Owner:** nobody


Currently the SMF created balanced procedures get the highest execLevel and are 
therefore executed last in the chain. There are cases where it is needed to 
execute procedures after the balanced procedures are completed. To offer this 
feature SMF can check which execLevel existing procedures have and inserting 
the balanced procedures one step after those.


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2240 dtm: Refactor the inter-node poll loop

[Anders Widell via Opensaf-tickets]

- **status**: review --> fixed
- **Comment**:

commit dea71bb783cc80d6dfb1fb6e761e86d9f7856d72 (HEAD -> develop, 
origin/develop)
Author: Anders Widell 
Date:   Wed Aug 16 11:50:30 2017 +0200

dtm: Support clusters with more than one hundred nodes [#2240]

Re-factor the inter-node poll loop by replacing the 100 entry fixed size 
poll
vector with a epoll event processing. This allows dtm to scale to clusters
larger than 100 nodes.




---

** [tickets:#2240] dtm: Refactor the inter-node poll loop**

**Status:** fixed
**Milestone:** 5.17.10
**Created:** Thu Dec 22, 2016 10:07 AM UTC by Anders Widell
**Last Updated:** Wed Aug 16, 2017 10:19 AM UTC
**Owner:** Anders Widell


Re-factor the DTM inter-node poll loop with the aim to increase scalability and 
performance. Today the poll loop has a fixed-length array with 103 positions, 
with puts an upper bound on the size of the cluster of 100 nodes. This 
limitation should be removed and the loop should be optimized to cope with 
larger clusters.


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2240 dtm: Refactor the inter-node poll loop

[Anders Widell via Opensaf-tickets]

- **status**: accepted --> review
- **Blocker**:  --> False
- **Milestone**: 5.17.08 --> 5.17.10



---

** [tickets:#2240] dtm: Refactor the inter-node poll loop**

**Status:** review
**Milestone:** 5.17.10
**Created:** Thu Dec 22, 2016 10:07 AM UTC by Anders Widell
**Last Updated:** Mon Apr 10, 2017 01:40 PM UTC
**Owner:** Anders Widell


Re-factor the DTM inter-node poll loop with the aim to increase scalability and 
performance. Today the poll loop has a fixed-length array with 103 positions, 
with puts an upper bound on the size of the cluster of 100 nodes. This 
limitation should be removed and the loop should be optimized to cope with 
larger clusters.


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2554 dtm: Re-factor the inter-node receive logic

[Anders Widell via Opensaf-tickets]

---

** [tickets:#2554] dtm: Re-factor the inter-node receive logic**

**Status:** assigned
**Milestone:** 5.17.10
**Created:** Wed Aug 16, 2017 09:00 AM UTC by Anders Widell
**Last Updated:** Wed Aug 16, 2017 09:00 AM UTC
**Owner:** Anders Widell


The DTM receive logic is unnecessarily complex: it calls recv() in many 
different places and has conditional statements that executes different pieces 
of code depending on how many bytes have been received on the stream socket so 
far. It should be refactored so that it uses only one single recv() statement 
and one single check to determine if we have received a full message or not.


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2491 imm: PBE regenerates imm.db if immnd exits during the PBE state verification

[Zoran Milinkovic via Opensaf-tickets]

- **status**: review --> fixed
- **Comment**:

develop:

commit deb8bea9703ef121c9c7324bcf6c94628fa0d7d1
Author: Zoran Milinkovic 
Date:   Wed Aug 16 10:53:58 2017 +0200

imm: regenerate PBE in verifyClassPBE only if database is corrupted [#2491]

In verifyClassPBE(), the patch makes distinguish between IMM issue and 
database corruption.
For IMM issue, PBE will not be regenerated, while for database corruption, 
PBE will be regenerated.

-

release:

commit e5052b27fd240d03344eb8a3249e1caf0d89c658
Author: Zoran Milinkovic 
Date:   Wed Aug 16 10:53:58 2017 +0200

imm: regenerate PBE in verifyClassPBE only if database is corrupted [#2491]

In verifyClassPBE(), the patch makes distinguish between IMM issue and 
database corruption.
For IMM issue, PBE will not be regenerated, while for database corruption, 
PBE will be regenerated.



---

** [tickets:#2491] imm: PBE regenerates imm.db if immnd exits during the PBE 
state verification**

**Status:** fixed
**Milestone:** 5.17.10
**Created:** Fri Jun 09, 2017 09:18 AM UTC by Zoran Milinkovic
**Last Updated:** Mon Aug 14, 2017 12:41 PM UTC
**Owner:** Zoran Milinkovic


If IMMND exits during the verification of PBE state, PBE regenerates new 
database from XML file.

PBE must distinguish between IMM issues (IMMND exits, network problem, etc) and 
database curruption.
If it's IMM issue, PBE should restart.
If it's database corruption issue, PBE should be regenerated.


2017-06-05T17:00:24.31 cm1 local0.notice osafimmnd[984]: NO This IMMND is now 
the NEW Coord
2017-06-05T17:00:24.31 cm1 user.info osafimmpbed: IN arg[0] == 
'/usr/lib64/opensaf/osafimmpbed'
2017-06-05T17:00:24.31 cm1 user.info osafimmpbed: IN arg[1] == '--recover'
2017-06-05T17:00:24.31 cm1 user.info osafimmpbed: IN arg[2] == '--pbe'
2017-06-05T17:00:24.31 cm1 user.info osafimmpbed: IN arg[3] == 
'/storage/clear/coremw/etc/imm.db'
2017-06-05T17:00:24.68 cm1 local0.err osafimmnd[984]: ER No IMMD service => 
cluster restart, exiting

2017-06-05T17:00:34.73 cm1 user.warning osafimmpbed: WA Verify class 
CmwMgntLockClass failed!
2017-06-05T17:00:34.73 cm1 user.notice osafimmpbed: NO Renamed 
/storage/clear/coremw/etc/imm.db to 
/storage/clear/coremw/etc/imm.db.failed_immdump because it has been detected to 
be corrupt.
2017-06-05T17:00:34.73 cm1 user.notice osafimmpbed: NO Removed obsolete journal 
file: /storage/clear/coremw/etc/imm.db-journal 
2017-06-05T17:00:34.73 cm1 user.warning osafimmpbed: WA verifyPbeState failed!
2017-06-05T17:00:34.73 cm1 user.warning osafimmpbed: WA Pbe: Failed to 
re-attach to db file /storage/clear/coremw/etc/imm.db - regenerating db file
2017-06-05T17:00:34.73 cm1 user.info osafimmpbed: IN Generating DB file from 
current IMM state. DB file: /storage/clear/coremw/etc/imm.db
2017-06-05T17:00:34.74 cm1 user.notice osafimmpbed: NO Successfully opened 
empty local sqlite pbe file /tmp/ImmPbeTmpSubDir/imm.db.d3HPkE



---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2553 dtm: Convert DTM to C++

[Anders Widell via Opensaf-tickets]

---

** [tickets:#2553] dtm: Convert DTM to C++**

**Status:** accepted
**Milestone:** 5.17.10
**Created:** Wed Aug 16, 2017 08:56 AM UTC by Anders Widell
**Last Updated:** Wed Aug 16, 2017 08:56 AM UTC
**Owner:** Anders Widell


To simplify the implementation of [#2551], convert DTM to C++. This will allow 
us to use STL containers instead of the more primitive patricia tree 
implementation in OpenSAF.



---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets

[tickets] [opensaf:tickets] #2544 imm: saClmDispatch is returning ERR_BAD_HANDLE until immnd initializes CLM handle

[Zoran Milinkovic via Opensaf-tickets]

- **status**: review --> fixed
- **Comment**:

develop:

commit 0b0d224ff3da7e59c7bb215664b1b39144a789f7
Author: Zoran Milinkovic 
Date:   Wed Aug 16 09:38:52 2017 +0200

imm: include CLM in poll before CLM handle is initialized [#2544]

CLM selection object is initially set to -1. Included CLM selection
object in poll will be ignored until CLM selection object is created
and set to fds[FD_CLM].

-

release:

commit 9f2329c6dd529e9196a2a2868ef251423e658330
Author: Zoran Milinkovic 
Date:   Wed Aug 16 09:38:52 2017 +0200

imm: include CLM in poll before CLM handle is initialized [#2544]

CLM selection object is initially set to -1. Included CLM selection
object in poll will be ignored until CLM selection object is created
and set to fds[FD_CLM].



---

** [tickets:#2544] imm: saClmDispatch is returning ERR_BAD_HANDLE until immnd 
initializes CLM handle**

**Status:** fixed
**Milestone:** 5.17.10
**Created:** Mon Aug 07, 2017 02:08 PM UTC by Zoran Milinkovic
**Last Updated:** Mon Aug 07, 2017 03:06 PM UTC
**Owner:** Zoran Milinkovic


In the poll loop, immnd handles FD_CLM event even if FD_CLM event is not 
processed by poll call.

With random values in fds[FD_CLM], immnd may process FD_CLM event without 
calling poll. saClmDispatch will return SA_AIS_ERR_BAD_HANDLE error until CLM 
handle is initialized and CLM selection object set to 
immnd_cb->clmSelectionObject.


---

Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net 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.--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot___
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets