Printers in printers list are discarded prematurely

[angelb]

Hello all.

When running CUPS in load-balance and fail-safe operation using an
alias for BrowsePoll, the client discard its printers list too soon
after it has no connection with a server.

client:
BrowseInterval 3600
BrowseTimeout 172800
BrowsePoll stlprcups01 <- alias in DNS
BrowsePoll stlprcups02 <- alias in DNS

After about 12 minutes, the client reported not being able to get
updates from server B.
E [09/Mar/2007:09:48:09 +0600] [cups-polld stlprcups02:631] CUPS-Get-printers failed: Unknown

Then, after about 90 minutes later from the time server B was stopped, the client started to discard all printers from server B:

-rw-r--r--   1 root     system       283439 Mar  9 10:08 printcap  <- original size
-rw-r--r--   1 root     system       173802 Mar  9 11:06 printcap  <- removal in progress
-rw-r--r--   1 root     system       162310 Mar  9 11:08 printcap  <- all printers from B are completely gone

On the client, I have BrowseTimeout set to 48-hrs(172800sec) but the
printers from server B were discarded in a hurry; 90 minutes after
server B was stopped.

I thought this may had something to do with using alias for BrowsePoll
but it was not the case. I've changed the BrowsePoll to use the real
host name but I got the same effect.

About 110 minutes after I stopped server B, the client started to
discard its printers list for server B.

I'm wondering if anyone has seen this before? I'm running cups 1.2.4.

Thanks,
Angel