Re:EPICS_CA_ADDR_LIST in R3.12Beta13CEBAF

[yamamoto@kekvax.kek.jp (Noboru Yamamoto)]

In >Message-Id: <9507171441.AA17610@aps.anl.gov>,
>From: Johnny Tang <tang@cebaf.gov>
wrote,
>Subject: CA-ADDR-LIST
>To: tech-talk@aps.anl.gov
>Date: Mon, 17 Jul 95 10:41:11 EDT
> 
>=============================================================
>Johnny Tang     Voice: (804)249-7239  E-Mail: tang@cebaf.gov
>Continous Electron Beam Accelerator Facility
>12000 Jefferson Avernue, MS 85A
>Newport News, Va 23606
>=============================================================
>Mailer: Elm [revision: 70.85]
> 
>> 
>> Hi,
>>  Does anyone use EPICS_CA_ADDR_LIST environment variable in R3.12Beta13?
>> I installed R3.12.0Beta13CEBAF on our HP-WS. ca_test cannot find a channel 
>> on IOC. The older version of ca_test , which also uses EPICS_CA_ADDR_LIST
>> variable , can find the channel. In our configuration, IOC's and OPI's 
>> belong differenct sub-nets and channel search by bradcast will not work.
>> 
>>  Thank you for your help,
>> 
>>   Noboru Yamamoto
>>   KEKB accelerator control group.
>>   KEK, JAPAN
>> 
> 
>  Since IOC's and OPI's in different sub-net in your configuration,
>you need set EPICS_CA_ADDR_LIST invironment variable. For more detail,
>refer "EPICS R3.12 Channel Access Reference Manual" which  you can
>get from APS WWW.
Yes. I set up "EPICS_CA_ADDR_LIST" but still ca_test in R312Beta13 could
not find the channel. ca_test from older version R312Beta0 can find the
channel when  "EPICS_CA_ADDR_LIST" is set up, however.

> 
>  R3.12.0Beta13CEBAF is CEBAF local customized (in devSup, recSup) version 
>of R3.12.0Beta13 from APS CVS server. They should be no difference in 
>channel access.
> 
>  --- Johnny Tang
>      CEBAF
> 
> 

In >Message-Id: <9507171541.AA02289@luke.atdiv.lanl.gov>
>From: hill@luke.atdiv.lanl.gov (Jeff Hill)
wrote,
> 
> 
>Hello,
> 
>I just sent a bug report to the EPICS exploder which is indirectly related
>to EPICS_CA_ADDR_LIST. I suspect however that this may not be your problem. 
>Here are a few things to check:
> 
>o Does the UNIX work station have a route to the IOC?
>       (type netstat -r under UNIX)
Yes. IOC and WS can talk to each other with TCP/IP.
 
>o Type "CASDEBUG=10" on the IOC. This will show too much detail about
>       the messages received by the CA server. However it may be
>       useful to see if your search requests are being processed by 
>       the server. Typing "CASDEBUG=0" will turn off the flood of 
>       messages.
When I used ca_test in R312Beta13 , no output was produced on IOC. 
Older ca_test  produced a lot.

>o Type "client_stat" on the IOC and check to see if the address list
>       used by the server matches the EPICS_CA_ADDR_LIST you have 
EPICS_CA_ADDR_LIST on IOC keeps just broad cast address of the sub-network
on which IOC's are connected. Shoud it includes IP addresses of the 
OPI's as well? 

>       specified. Watch the output from ioc_init for possible syntax
>       errors in the resource definition file.


>o Use a network sniffer such as etherfind or tcpdump. I often find 
>       that the command "etherfind -proto udp dstport 5064"
>       provides useful information about who is sending search 
>       requests (5064 is the default port for UDP CA search requests).
>       Note that it is best to use a different machine than the
>       one running the client that you are observing.
I don't have both of them on our HP-WS. I'm going to install them.

> 
>That is all I can think of at the moment. If you are still having 
>problems let me know and I will examine your situation in more
>detail.
> 
>Jeff
> 

Thank both of you for your help.

Noboru