Re: autosave failure S_rpcLib_RPC_CANTSEND

[Ron Sluiter <sluiter@aps.anl.gov>]

On 11/8/2013 5:46 PM, J. Lewis Muir wrote:
> Well, not so fast I guess.  It just happened again on Saturday, this
> time using NFSv2 (using the image you sent me).  The error messages on
> the console look the same:
>
> === console ===
> save_restore:write_it: fprintf returned -1. [131101-194540]
> ../save_restore.c(1636): [0x2f0003]=write_it:S_rpcLib_RPC_CANTSEND
> save_restore:write_it: fclose('auto_settings.sav') returned -1
> ../save_restore.c(1751): [0x2f0003]=write_it:S_rpcLib_RPC_CANTSEND
> save_restore:write_it: Giving up on this attempt to write 'auto_settings.sav'. [131101-194540]
> ===============
>
> Any ideas on how to investigate, fix, or work around?

Hello Lewis,

I mentioned this before in the tech-talk thread, but you may have missed it
amongst a barrage of questions.

We had sever problems with both save/restore and corrupted saveData 
*.mda files
at several 2-ID-D IOC's this summer.  A sampling of error messages,

> save_restore:write_save_file: Backup file 
> (/xorApps/epics/synApps_5_6/ioc/2idd/iocBoot/ioc2idd/autosave/auto_settings.savB) 
> bad or not found. Writing a new one.
> [130723-124613] myFileCopy: size=32768, chars_printed=15457
> save_restore:write_it - unable to open file 
> '/xorApps/epics/synApps_5_6/ioc/2idd/iocBoot/ioc2idd/autosave/auto_settings.savB'
> [130723-124613] ../save_restore.c(1346): 
> [0x2f0003]=write_it:S_rpcLib_RPC_CANTSEND
> ---- splice here -----
> saveData: ...will retry in 5 seconds
> saveData: ...will retry in 5 seconds
> saveData: * saveData: too many retries; abandoning data from scan 
> '2idd:scanH'
> saveData: * saveData:proc_scan_cpt(2idd:scan1): can't open data file!!

I tried a number of solutions. Since we were running VxWorks 6.8 when the
problems started I tried switching back to VxWork 5.5.2; still failed.

I switched back to VxWorks 6.8 and tried using NFSv2; still failed.

The problems were resolved (so far) by upgrading to VxWorks 6.9.
We never found the root cause (which I find very annoying).

(I did create a a procedure that provided nearly 100% failure rate for 
corrupted
*.mda files from saveData. Contact me if you want more info on this.)

Ron