nikhilgupta10
commented
7 years ago nikhilgupta10 imported these comments from Sourceforge: "jcfuller":Additional note: When running this on Ubuntu 12.1 64-bit, the model started at 1 GB of RAM. After 2 months of simulation it was a 2 GB of RAM. It appears to keep growing in size (and slowing down the machine).
,
"jcfuller":Additional note: The \memory leak\ appears in Windows 64-bit also. These may be unrelated issues.
,
"ciraci": * status changed from new to accepted
I found a leak in climate module. That seems to have fixed the issues with one of the failing models (I didn't do a long run, e.g. >2days). The other one didn't fail in ubuntu 64-bit. So these might be two unrelated errors.
One of the models causes minor memory corruptions in core. I'm fixing them now.
,
"ciraci":Fixed another leak in curver.cpp. Might solve the problem with one of the test models.
,
"ciraci": * owner changed from ciraci to andyfisher
- status changed from accepted to assigned
I think memory issues are fixed. Assigning it to Andy for validation.
,
"mactyr":Thank you all for your work on the memory usage issue! I'm the external user who reported both of the issues mentioned in this ticket. I wanted to add that I have been able to replicate the bad logging (the \gibberish\ readings) on Windows 7 (64 bit) as well, using the same models that I sent to Jason.
,
"andyfisher": * status changed from assigned to closed
- resolution set to fixed
fixed and committed for patch 2.3.1
,
External user reports that model starts to create \gibberish\ is CSV output well into simulation. In some instances, the model fails out with a \kill\ (i.e., it doesn't exit cleanly through a GLD exit command). Both Frank and Jason can replicate a similar error, however, it occurs at different times and only in Linux versions. Frank's fails in May, Jason's in January, and user's in February. This may be linked to other reports by the same user of massive memory leakage within the models.
Contact Jason for sample model that fails.
,