RMS shared write read inconsistensy

[xyzard]

We are experiencing problems with finding a newly written record (RMS shared 2 Gb index file) followed by a direct, mailbox initiated, read in an other process (can't find the record). The record is later accessible. This problem has never occured before but araised after an upgrade from AS800 VMS 6.2 to DS15 7.3-2 (with plenty of memory 2G and aprox max 1 Gb reserved to buffers)
The VMS-system is newly generated (from scratch) with default (no exeptional) sysgen parameters. Volume shadowing is enabled on all disks. Any ideas?
Regards
Bo

[labadie_1]

You have applied the latest patches (Rms and others) ?

[Ian Miller.]

One process writes a record, send a message via mailbox to another process, the other process reads the record and fails - is this correct?

How does the other process know which record to use i.e. how is it specifing which record to read?

What is the error status returned when trying to read the record?

What are the sharing options specified for the file?

[Wim Van den Wyngaert]

Once saw an analogue problem. The program first did the mailbox write and then the write to the file (oracle db). Worked on VAX but when moved to much faster alpha it failed.

Wim

[xyzard]

Thanks Labadie and Ian for Your interest!!

The problem is somtetimes not what it seems to be.

VMS is still solid as a rock!

Regards
Bo

[xyzard]

Thanks to all!
Regards Bo

[Wim Van den Wyngaert]

It would help others if you explained what was the problem.

[xyzard]

It was by accident two active database version missmatched by a failing backuproutine and the old db was held by the reading process. Of course it would be asking to much for VMS to handle this situation also reading records in the newer db :-). Sorry to occupy your time.
Again thanks to all!!
Regards
Bo