cmcld: Changed serial device to RECOVERING

[Rudy Williams]

Folks--

I am seeing the syslog message in the subject very often on the two nodes in a cluster. I do indeed have a null modem connecting the aux ports on these rp4440s. One second later, the message "cmcld: Changed serial device to UP" appears in the syslog.

Haven't seen this one before. How can I trouble shoot this problem? Might the timeout values be set too low?
Thanks.

[spex]

Hi Rudy,

Look half-way down on this page:

http://docs.hp.com/en/B3935-90012/ch01s03.html

PCS

[Prashanth.D.S]

Hi Rudy,

The daemon updates the status database whenever it sees the
serial port change state. At the time of the update, the daemon
writes this message:

cl_log(LOG_EXTERNAL|0, RCM, "Changed serial device status to
%d\n", s_status.status);

The port can have these statuses:

Number Status
------ ------
0 UNINITIALIZED
1 DOWN
2 UP
3 TIMEOUT

If a read or write to a port gets an error, the status is changed to DOWN and the status database updated. Once it reads or writes successfully, it is changed to UP. TIMEOUT is rarely seen.


Best Regards,
Prashanth

[Rudy Williams]

PCS--The error that I am seeing does not appear on the page that you noted.


Prashanth--This appears to be on track with what I am seeing. However, I do not see "RECOVERING" as a status entry.

I tightened the connectors on the serial cable. If this does not help, I'll try another cable.

Thanks.

[Prashanth.D.S]

Hi Rudy,

This usually means that the RS232 connection (the serial link) was marked down
at one point, and now it is coming back up..

Basically if the system is busy, the header inform doesn't get through on the first attempt,although later it does get through.
In these cases, instead of marking the RS232 link "DOWN", it is marked "Recovering" (an interum state)

Do you have the hearbeat lan in the same subnet ?

Best Regards,
Prashanth

[Prashanth.D.S]

I mean do you have have multiple heartbeat subnets ??

[Rudy Williams]

Prashanth--

Yes, I have two subnets for the heartbeat.

PRIMARY lan0 (crossover cable to other node - private subnet 192.168.3.0)
PRIMARY lan1 (to Ethernet swtich - data subnet 10.33.56.0)
STANDBY lan2 (standby for lan1)

The servers really aren't that busy all of the time, they run batch-oriented software. So, they are either very idle or very busy. I see these messages all the time.

Rudy

[melvyn burnard · Accepted solution]

If you already have two LANs configured as HERATBEAT_IP then you should remove the serial heartbeat.
It should only be used when there is one LAN connection between servers.

[Rudy Williams]

Folks--

We have MC/SG 11.16 in this cluster. It was stated that the serial heartbeat should not be used when two LAN heartbeats are in place.

The documentation for version 11.16 backs this up. We removed the serial device. Thanks!