Preeti,
I find your idea about oh_evtget_thread_loop quite appealing.
As for oh_evtpop_thread loop cannot agree. Without this thread you will
see no events at all.
Discovery thread need some investigation. There can be a plug-in that
relies on it.
And discovery thread is woken up when HPI application calls
saHpiDiscover().
About OA threads I cannot say anything.
Anton Pak
On Tue, 21 Jun 2011 20:35:40 +0400, Preeti Sharma
<[email protected]> wrote:
> Hi Anton,
> I am running openhpid in gdb. I set break point at oh_process_events
> (this method is called in oh_evtpop_thread loop). I ran client hpipower
> to generate event once openhpid initialization and discovery completed.
> I am not seeing any hit to my break point. I see hit to this break point
> only during discovery. Hope this is safe for me to comment out
> oh_evtget_thread_loop creation to reduce processing load. I am as well
> interested to comment out oh_evtpop_thread loop (if it is not needed).
> As well, why we should have discovery thread running after every 3
> minute (once initialization of plugins is complete), it is to handle
> clients' saHpiDiscover calls?
>
> As well I am eager to know out of 5 threads which threads are per OA
> basis.
>
> Regards,
> Preeti
>
> -----Original Message-----
> From: Anton Pak [mailto:[email protected]]
> Sent: Tuesday, June 21, 2011 11:55 AM
> To: Chinnaswamy, Elangoraja; Preeti Sharma;
> [email protected]
> Subject: Re: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>
> Preeti,
>
> oh_evtpop_thread serves OpenHPI event queue.
> It wakes when the queue is not empty.
>
> oh_evtget_thread_loop mission is not clear now.
> It periodically calls get_event function for every handler.
> As a result of the get_event function we can get one or several events
> added to OpenHPI event queue.
> May be in the past there were plug-ins that couldn't collect events
> themselves and needed periodic driving on from
> the OpenHPI daemon.
>
> Anton Pak
>
> On Tue, 21 Jun 2011 19:18:55 +0400, Preeti Sharma
> <[email protected]> wrote:
>
>> Hi Elangoraja,
>> The messages as you had guessed- were due to HEART_BEAT. I am
>> summarizing my understanding below:
>>
>> When we start openhpid - three daemon threads are started (each one as
>> described below):
>>
>> * Oh_discovery_thread_loop - This calls discover_resources (via
>> a call to oh_discovery) on plugin handlers (as we define in
>> openhpi.conf). discover_resources of plugin's initializes plugin. This
>> happens every three minute though once plugin is initialized further
>> call to discover_resources of specific plugin just return. Here
>> extensive messaging happens (login to OA and get all resources
>> information).
>>
>> * oh_evtget_thread_loop
>>
>> * oh_evtget_thread_loop
>>
>>
>>
>>
>>
>> OA_SOAP plugin threads:
>>
>> Two threads(oa_soap_event_thread) are created one for active OA and
>> another for standby. These threads are created from discover_resources
>> during plugin initialization. oa_soap_event_thread method waits till
>> discovery get over. This is the thread which is sending event request to
>> OA and getting response back. When there are no event for the OA to sent
>> it send HEATBEAT_EVENT. In case there is an event it is processed and
>> event is added to Daemon's event queue. Client program read events from
>> this queue.
>>
>>
>> What is the role of these two threads oh_evtget_thread_loop,
>> oh_evtget_thread_loop? Do they cause any messaging to happen? Are they
>> significant for OA_SOAP plugin or useful for other plugins? What is
>> "SIGNALED: Got signal from plugin" means in these methods as well in
>> oh_discovery_thread_loop?
>>
>>
>>
>>
>>
>> Regards,
>>
>> Preeti
>>
>>
>> From: Chinnaswamy, Elangoraja [mailto:[email protected]]
>> Sent: Monday, June 20, 2011 3:15 PM
>> To: Preeti Sharma
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> Hi Preeti,
>> This may be that the OA is not accessible. Check if you can login to the
>> OA's successfully using web interface.
>> One more thing that you should consider is the HEART_BEAT event that is
>> generated from OA. This should be more frequent.
>>
>> Thanks,
>> Elango
>>
>> From: Preeti Sharma [mailto:[email protected]]
>> Sent: Friday, June 17, 2011 2:51 PM
>> To: Chinnaswamy, Elangoraja
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> Hi Elangoraja,
>> I see in /var/log/messages below logs (I am running openhpid with -v
>> option):
>>
>> Jun 17 16:43:22 irc1 lt-openhpid: ERROR: (oh_ssl.c, 466, SSL error: bad
>> hostname lookup)
>> Jun 17 16:43:24 irc1 lt-openhpid: ERROR: (oh_ssl.c, 464,
>> BIO_do_connect() failed)
>> Jun 17 16:43:24 irc1 lt-openhpid: ERROR: (oh_ssl.c, 466, SSL error: bad
>> hostname lookup)
>> Jun 17 16:43:26 irc1 lt-openhpid: ERROR: (oh_ssl.c, 464,
>> BIO_do_connect() failed)
>> Jun 17 16:43:26 irc1 lt-openhpid: ERROR: (oh_ssl.c, 466, SSL error: bad
>> hostname lookup)
>> Jun 17 16:43:28 irc1 lt-openhpid: ERROR: (oh_ssl.c, 464,
>> BIO_do_connect() failed)
>> Jun 17 16:43:28 irc1 lt-openhpid: ERROR: (oh_ssl.c, 466, SSL error: bad
>> hostname lookup)
>> Jun 17 16:43:30 irc1 lt-openhpid: ERROR: (oh_ssl.c, 464,
>> BIO_do_connect() failed)
>> Jun 17 16:43:30 irc1 lt-openhpid: ERROR: (oh_ssl.c, 466, SSL error: bad
>> hostname lookup)
>> Jun 17 16:43:32 irc1 lt-openhpid: ERROR: (oh_ssl.c, 464,
>> BIO_do_connect() failed)
>> Jun 17 16:43:32 irc1 lt-openhpid: ERROR: (oh_ssl.c, 466, SSL error: bad
>> hostname lookup)
>>
>> Regards,
>> Preeti
>> From: Chinnaswamy, Elangoraja [mailto:[email protected]]
>> Sent: Friday, June 17, 2011 3:25 PM
>> To: Preeti Sharma
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> Hi Preeti,
>> It will be helpful if you can share the log files (using -v option),
>> can you share the /var/log/messages file that was captured during this
>> tcpdump? That will help. Did you see any retries for establishing
>> connection with OA?
>>
>> Thanks,
>> Elango
>>
>> From: Preeti Sharma [mailto:[email protected]]
>> Sent: Thursday, June 16, 2011 4:39 PM
>> To: Chinnaswamy, Elangoraja
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> Hi Elangoraja,
>> Thanks for clarification. I am running only openhpid (no openhpi clients
>> running). I see continuous messages in TCPDUMP coming and going to OA
>> and these messages appear very frequent than 3 minutes, what are these:
>> (I am pasting below logs)
>>
>> [root@irc1 etc]# tcpdump -i eth0 -vv ip6 src <ipv6 address of OA>
>> tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 96
>> bytes
>> 18:33:37.987988 IP6 (hlim 63, next-header: TCP (6), length: 40)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: S, cksum
>> 0x1da7 (correct), 2030375816:2030375816(0) ack 64742404 win 5712 <mss
>> 1440,nop,nop,timestamp 2277318774 472366580,nop,wscale 6>
>> 18:33:37.988276 IP6 (hlim 63, next-header: TCP (6), length: 32)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: ., cksum
>> 0x5e70 (correct), 1:1(0) ack 212 win 106 <nop,nop,timestamp 2277318774
>> 472366580>
>> 18:33:38.107543 IP6 (hlim 63, next-header: TCP (6), length: 1184)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: P
>> 1:1153(1152) ack 212 win 106 <nop,nop,timestamp 2277318804 472366580>
>> 18:33:38.114550 IP6 (hlim 63, next-header: TCP (6), length: 32)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: ., cksum
>> 0x587b (correct), 1153:1153(0) ack 410 win 123 <nop,nop,timestamp
>> 2277318806 472366706>
>> 18:33:38.203905 IP6 (hlim 63, next-header: TCP (6), length: 91)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: P
>> 1153:1212(59) ack 410 win 123 <nop,nop,timestamp 2277318829 472366706>
>> 18:33:38.204545 IP6 (hlim 63, next-header: TCP (6), length: 32)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: ., cksum
>> 0x5729 (correct), 1212:1212(0) ack 559 win 140 <nop,nop,timestamp
>> 2277318829 472366796>
>> 18:33:38.204868 IP6 (hlim 63, next-header: TCP (6), length: 32)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: ., cksum
>> 0x54b0 (correct), 1212:1212(0) ack 1172 win 159 <nop,nop,timestamp
>> 2277318829 472366797>
>> 18:33:39.047437 IP6 (hlim 63, next-header: TCP (6), length: 970)
>> jupcutoa1av6.https > 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.52594: P
>> 1212:2150(938) ack 1172 win 159
>>
>>
>>
>> tcpdump -i eth0 -vv ip6 dst <ipv6 address of OA>
>> tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 96
>> bytes
>> 18:36:02.909204 IP6 (hlim 64, next-header: TCP (6), length: 181)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: P
>> 219776186:219776335(149) ack 2183677272 win 63 <nop,nop,timestamp
>> 472511501 2277355370>
>> 18:36:02.909496 IP6 (hlim 64, next-header: TCP (6), length: 821)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: P
>> 149:938(789) ack 1 win 63 <nop,nop,timestamp 472511502 2277355370>
>> 18:36:03.128248 IP6 (hlim 64, next-header: TCP (6), length: 32)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: ., cksum
>> 0x9411 (correct), 938:938(0) ack 2857 win 86 <nop,nop,timestamp
>> 472511720 2277355425>
>> 18:36:03.130181 IP6 (hlim 64, next-header: TCP (6), length: 32)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: ., cksum
>> 0x9242 (correct), 938:938(0) ack 3296 win 108 <nop,nop,timestamp
>> 472511722 2277355425>
>> 18:36:03.130226 IP6 (hlim 64, next-header: TCP (6), length: 69)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: P
>> 938:975(37) ack 3296 win 108 <nop,nop,timestamp 472511722 2277355425>
>> 18:36:03.130272 IP6 (hlim 64, next-header: TCP (6), length: 32)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: F, cksum
>> 0x921c (correct), 975:975(0) ack 3296 win 108 <nop,nop,timestamp
>> 472511722 2277355425>
>> 18:36:03.130840 IP6 (hlim 64, next-header: TCP (6), length: 32)
>> 2001:5b0:ffff:fff4:217:a4ff:fe77:2a.57054 > jupcutoa1av6.https: ., cksum
>> 0x9219 (correct), 976:976(0) ack 3297 win 108 <nop,nop,timestamp
>> 472511723 2277355426>
>>
>> 7 packets captured
>> 7 packets received by filter
>> 0 packets dropped by kernel
>>
>> Regards,
>> Preeti
>>
>> From: Chinnaswamy, Elangoraja [mailto:[email protected]]
>> Sent: Thursday, June 16, 2011 3:20 PM
>> To: Preeti Sharma
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> Hi Preeti, yes; minor correction below.
>> The daemon initialization will take some time depending on the resources
>> found in the enclosure, anytime between few seconds to few minutes. The
>> OpenHPI framework ABI calls (discovery)are executed every 3 minutes till
>> the openhpid is killed/shutdown; not only during initialization.
>>
>> You can also execute the openhpid using -v option to see what is going
>> on, if that helps.
>>
>> Thanks,
>> Elango
>>
>> From: Preeti Sharma [mailto:[email protected]]
>> Sent: Thursday, June 16, 2011 11:56 AM
>> To: Chinnaswamy, Elangoraja
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>>
>> Hi Elangoraja,
>> Just confirming, what I understood (is this correct?):
>> During Daemon initialization OpenHPI framework calls discover ABI every
>> 3 minutes. Once initialization is done, OAs sends event for any changes
>> and OAs specific thread in Daemon handles it. How much time Daemon takes
>> to initialize?
>>
>> I came across method oh_evtget_thread_loop timer:
>>
>> #define OH_EVTGET_THREAD_SLEEP_TIME 3 * G_USEC_PER_SEC
>>
>> Looks like this method is checking for OA events after every three
>> second and passing this to plugin handler
>>
>> Regards,
>> Preeti
>> From: Chinnaswamy, Elangoraja [mailto:[email protected]]
>> Sent: Thursday, June 16, 2011 1:23 PM
>> To: Preeti Sharma
>> Cc: [email protected]
>> Subject: RE: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> Hi Preeti,
>> As said, the OpenHPI framework calls the discover ABI every 3 minutes.
>> The oa_soap plug-in returns a success, instead of doing a re_discover on
>> the OA's. This holds good as long as the initial discovery (after the
>> plug-in is initialized) is completed successfully once. The subsequent
>> changes in the resources are handled by events.
>>
>> This should clarify, let me know.
>>
>> Thanks,
>> Elango
>>
>> From: Preeti Sharma [mailto:[email protected]]
>> Sent: Thursday, June 16, 2011 11:16 AM
>> To: Chinnaswamy, Elangoraja
>> Cc: [email protected]
>> Subject: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>>
>> Hi Elangoraja,
>>
>> Thanks, for the response. Just I am not clear what below line means (if
>> you can please clarify it further):
>>
>> "the oa_soap plugin ignores and returns success if the discover is
>> complete after the initial discovery."
>>
>>
>>
>> Regards,
>>
>> Preeti
>>
>>
>>
>> Date: Tue, 14 Jun 2011 17:14:05 +0000
>>
>> From: "Chinnaswamy, Elangoraja" <[email protected]>
>>
>> Subject: Re: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>> To: "[email protected]"
>>
>> <[email protected]>
>>
>> Message-ID:
>>
>>
>> <1be8d981ba123b47a1d153beb20d1fbbb46ca07...@gvw1087exb.americas.hpqcorp.net>
>>
>>
>>
>> Content-Type: text/plain; charset="us-ascii"
>>
>>
>>
>> Hi Preeti,
>>
>> Without any openhpi client running. At first the 'discovery' takes place
>> on the OA (active and standby). The OpenHPI framework executes the
>> discover ABI every 3 minutes, the oa_soap plugin ignores and returns
>> success if the discover is complete after the initial discovery. After
>> this any change to the resources are handled by the event thread (one
>> thread for each OA). Now the traffic between the oa_soap plugin and the
>> OA is only when an event occurs in the OA (change in resources).
>>
>>
>>
>> Hope this clarifies?
>>
>>
>>
>> Thanks,
>>
>> Elango
>>
>>
>>
>> -----Original Message-----
>>
>> From: Preeti Sharma [mailto:[email protected]]
>>
>> Sent: Monday, June 13, 2011 2:29 PM
>>
>> To: [email protected]
>>
>> Subject: [Openhpi-devel] Traffic bet openhpi daemon and OA??
>>
>>
>>
>> Hi,
>>
>> I want to know detail about continuous traffic between openhpi daemon
>> and OA (without any openhpi clients running). What messages are
>> exchanged? Who is sender and who responds...? Or basically when a Deamon
>> is started what interaction happens between daemon and OA initially, and
>> at regular intervals.
>>
>>
>>
>> Regards,
>>
>> Preeti
------------------------------------------------------------------------------
EditLive Enterprise is the world's most technically advanced content
authoring tool. Experience the power of Track Changes, Inline Image
Editing and ensure content is compliant with Accessibility Checking.
http://p.sf.net/sfu/ephox-dev2dev
_______________________________________________
Openhpi-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openhpi-devel
