TCL question on max file size

[Tom Patton]

Sorry, this is not specifically an engine question….

I’m using hcitcl from our 5.3 UNIX environment and found that when I do some batch type processing with a file that is over 38 or 39 Meg, I get the following error:

/hci/TEST/bin/meditech2conv1[86]: 37188 IOT/Abort trap(coredump)

and a 217 Meg core dump.

If I split the file into smaller pieces (under 38 Meg) the files process just fine.

Is there a size limitation I’m bumping into?

Topic

[Nathan Martin]

What does “ulimit -a” say?

[Tom Patton]

Nathan, thanks for the response ulimit say:

$ ulimit -a

time(seconds)        unlimited

file(blocks)         2097151

data(kbytes)         393216

stack(kbytes)        16384

memory(kbytes)       98304

coredump(blocks)     2097151

nofiles(descriptors) 2048

[Nathan Martin]

I don’t think it’s a TCL limit.  I think it’s failing to get as much memory as it wants.  (Although, it seems like the core file would be closer to 393216 kbytes — maybe you’re running out of drive space for the core, or maybe TCL’s buffer growth algorithm grows exponentially.)

I’d make a new script that only reads and processes a few records at a time.  Look for multiple copies of large vars and clear those out, too.  (set original_file_data {})

You could also try to change your ulimit(s) while your logged in (up to any hard limits) and run the script again, but I wouldn’t recommend it without checking available memory/swap/AIX admin attitude.  Other programs could start to run out of memory and bad things could happen.

[Author]

Replies