27-Oct-2012

Reboot required
Trying to copy a number of files using WSFTP – a Windows FTP client – Diana hang for some reason, and even the DecWindows terminal did not respond – didn’t wake up from the sleeping mode it was in: the screensaver didn’t respond at all. It happened when I try to enter a location that is shared using Samba – so I stopped the WSFTP session on the PC and rebooted, to be sure all connections were really lost. But that didn’t help, so I tried the alternate way: starting Daphne – but I lost the password so I had to startup with UAFALTERNATE set to 1 and STARTUP_P1 to “MIN” to set the password, so I could enter a new one. In that process, Diana suddenly woke up – the processes locking the machine seemed to time out., the screensaver woke up and I was able to login. Nevertheless, I kept working on Daphne to be sure to have a second way into Diana.
Using Accounting I could only find that some FTP-processed were stopped on forced exit.
So I retried the WSFTP session, and found that there was one process running into FPG status = free pages were exhausted, the program disappeared after some time, another such process was started and ran into the same problem – and so on. At some time, I was able to delete such a process, but soon another appeared, and was left in an MWAIT state. Stopped that one, which caused WSFTP to loose connection.
In nitself, there is nothing wrong, but it started when I changed directory in WSFTP, to a location that is shared in Samba. So I stopped Samba, (There are still a number of things to find out about it: because I couldn’t login from the PC at some time – probably because the password was invalid? And resetting it in Samba failed…) but that didn’t make a difference: Retried WSFTP sequence again, but now the system hung again – and became inresponsive. Even from Daphne I couldn’t connect ($ MC SYSMAN, then SET ENVIRONMENT/NODE=DIANA and DO SHO SYS/PROC=*FTP*) – hanging that node as well (I could sissue an interrupt but ANY action next would hang the that node as well.
There was one alternative: reset Diana and boot….
But what causes this FPG state? I googled it and found it might be caused by too little free pagefile space, or a badly fragmented pagefile. Not on Diana, I think, because I have more pagefile space than I have internal memory:

Swap File Usage (8KB pages):                   Index        Free        Size
 DISK$AXP083:[SYS0.SYSEXE]SWAPFILE.SYS
                                                  1        1304        1432

 Total size of all paging files:                                     195800
 Total committed paging file usage:                                   37316


The pagefile on AXP083: seems to be contiguous, if I interpret DIR/FULL correctly. I could do an ANA/RMS/STAT on that file but it didn’t reach the end in a sufficient short time. But I cannot examine the files on ALPHASYS:[LOCAL]. Not directly because these are locked, so I dumped INDEXF.SYS and looked to the number of retrieval pointers, and I found that all pagefiles on ALPHASYS:[LOCAL] are fragmented, but not that bad: the worst fragmented is Pagefile.sys, but all fit within the same header file… nevertheless, it could be useful to re-create pagefiles – and with less, or no , fragmentation….
On the other hand: straight FTP is not giving any trouble, so it may have to do with WSFTP itself….