I ran truss on a running java program, and noticed that the access system call was returning "Err#2 ENOENT" for existing files, files that were succesfully "accessed" earlier in the same program, and later on.


Is there any reasonable explanation for this behaviour?



In the truss output I find: (preceded by line-numbers)

+552 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 0
+563 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 0
+576 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 1
+607 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 0
+1234 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 66
+1384 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) Err#2 ENOENT
+1452 access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 0


The files are located on:
/dev/lv_soft_IH /softih jfs2 Aug 06 06:37 rw,log=INLINE

Any ideas are very well come,

Gerard

Is the page dynamic? could it have been in the process of being updated while you / your process was trying to read it?

ENOENT 2 /* No such file or directory */

This suggests the file was not there at the time.

Or are there any errors for this disk / mount at the time of the problem?

Otherwise I can not think of a rational explanation.

access("/softih/fatwire/Shared/elements/Roularta/Query/Template/Sql/2_0.jsp", 04) = 66


Do you have a clue what the 66 would indicate?

I can hardly imagine that filesystem errors would go unnoticed by the system administrators, but I'll have a look anyhow. Thanks.

---------- Post updated at 08:36 PM ---------- Previous update was at 06:31 PM ----------

What if we did not have enough aioserver processes?

It looks that way.

Sorry, no idea.
I suspect, like you, google can only find your threads and unassociated rubbish.
Could it be a line number in the .jsp???

Hi,

there is a known bug that sounds similar to yours ... according to IBM it's NFS related but maybe you want to have a look:
Kind regards
zxmaus

Thanks, zxmaus. I'd ignored that one because of the NFS, and also the files in my case have been around for months without change, to the file and all directories above.

The filesystem is used as a disk cache by a CMS called Fatwire, loads of creates and deletes of files.

I wonder if it is not the truss output that is having me on, eg:

4776365: 5.3595: statx("/softih/HTMLpages/OC4J_Fatwire/application-deployments/ContentServer/cs/persistence/_pages/_jsp/_cs__deployed/_roularta/_layout/_Logic.class", 0x000000011749FA40, 176, 010) Err#2 ENOENT
= 0

Has this ENOENT got anything to do with this statx? What's the "= 0" then?