- **status**: review --> fixed
- **Comment**:
commit f6f70b623b202d9bba2e53156531104ef772754b (HEAD -> develop,
origin/develop)
Author: thuan.tran <[email protected]>
Date: Thu Aug 27 13:13:01 2020 +0700
mds: improve mdstest suite 27 [#3216]
- Update tet_receiver() to poll without timeout as sender may
take long time for sendto() return due to run out of memory.
- Update tet_sender() to slow down sending if amount of message
is big and message size is big to avoid kernel kill it as memory
usage too much.
commit bb625e162390e8e83f18599da68a3401238db795
Author: thuan.tran <[email protected]>
Date: Wed Aug 19 15:14:56 2020 +0700
mds: fix receiving old msg under flow control enabled [#3216]
- Revert apart of #3151 solution, not decide PortId reset base on
fseq=1 but reset rcvwnd when getting Intro msg from known PortId.
- Check to skip invalid Nack to avoid sender mistake move to
overflow and queue all messages later but receiver don't get any
further message to send ChunkAck.
- Not return error if PortId not found in checking send queue
capable to avoid agent crash after fix #3208 if agent enable mds
flow control.
---
** [tickets:#3216] mds: receiving old msg under flow control enabled**
**Status:** fixed
**Milestone:** 5.20.08
**Created:** Mon Aug 24, 2020 04:56 AM UTC by Thuan Tran
**Last Updated:** Mon Aug 24, 2020 06:27 AM UTC
**Owner:** Thuan Tran
mdstest 27 (test 7 and 8) sometimes fail as following
~~~
Received Message len = 1000
The message is=65537
MDS RETRIEVE is SUCCESSFULL
Got 66349 message from 1c60c7c3
Expect 66349 but received 65537 from 1c60c7c3
~~~
Due to unexpected retransmission go through boundary value and receiver mistake
see as [portid reset on sender]
Receiver accept that old message then test fail, sender also detect msg
disorder base on chunk ack from receiver.
~~~
<139>1 2020-08-18T06:53:55.265043+02:00 SC-1 mdstest 3743 mds.log [meta
sequenceId="124"] FCTRL: [me] <-- [node:1001001, ref:476104643],
RcvData[mseq:65535, mfrag:0, fseq:65535], rcvwnd[acked:811, rcv:812, nacked:0],
Error[unexpected retransmission]
<139>1 2020-08-18T06:53:55.265049+02:00 SC-1 mdstest 3743 mds.log [meta
sequenceId="125"] FCTRL: [me] <-- [node:1001001, ref:476104643],
RcvData[mseq:65536, mfrag:0, fseq:0], rcvwnd[acked:811, rcv:812, nacked:0],
Error[unexpected retransmission]
<141>1 2020-08-18T06:53:55.265054+02:00 SC-1 mdstest 3743 mds.log [meta
sequenceId="126"] FCTRL: [me] <-- [node:1001001, ref:476104643],
RcvData[mseq:65537, mfrag:0, fseq:1], rcvwnd[acked:811, rcv:812, nacked:0],
Warning[portid reset on sender]
...
<139>1 2020-08-18T06:53:55.265128+02:00 SC-1 mdstest 3748 mds.log [meta
sequenceId="1143"] FCTRL: [me] <-- [node:1001001, ref:3809107711],
RcvChkAck[fseq:1, chunk:1], sndwnd[acked:811, send:1000, nacked:196084],
queue[size:188], Error[msg disordered]
<139>1 2020-08-18T06:53:55.265141+02:00 SC-1 mdstest 3748 mds.log [meta
sequenceId="1144"] FCTRL: [me] <-- [node:1001001, ref:3809107711],
RcvChkAck[fseq:4, chunk:3], sndwnd[acked:811, send:1000, nacked:196084],
queue[size:188], Error[msg disordered]
~~~
---
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
