GnuCash 2.6.3 (portable) for Windows taking a LONG time to read/save file

[John Ralls]

On Jun 29, 2014, at 9:15 PM, Wm <wm+gnc at tarrcity.demon.co.uk> wrote:

> Sun, 29 Jun 2014 08:37:18 <6BD749E4-9734-44CE-95A5-2B84120C1969 at ceridwen.us> John Ralls <jralls at ceridwen.us>
> 
>> 
>> On Jun 28, 2014, at 3:46 PM, Lance Alligood <lalligood at gmail.com> wrote:
> 
> [snip]
> 
> Hi, L, you've provided some details I didn't have to work with before <fx: wry grin>.  You're certainly in good hands now.
> 
>> I looked over the conversation on the Portable Apps site, and I see that you already tested without compression and with the SQL backend. The compression overhead seems high to me, but I haven't ever profiled it so I don't know whether that's typical or not.
> 
> My feeling at the moment is that it may be a symptom or artefact.
> 
>> That block of several thousand repetitions of "attempt to write disabled transaction log" messages looks like the meat of the problem. since it appears to start 8 seconds after program start and the next segment that you indicate is 18 minutes later.
> 
> Agreed, see below.
> 
>> Since you say it takes only 2-3 minutes to load your file, it doesn't seem likely to me that those are going on for all 18 minutes. Is that correct? If so, what is the actual duration?
> 
> More to the point why the time difference between the *nix and Win systems with the same file?
> 
>> Can you compare the log information of a windows and a linux run of GC? Is there really nothing else being logged in that segment of the log file?
> 
> That seems important now.
> 
> L, now we're on equal ground a simple test might be:
> 
> run gnc with  --debug --extra on both platforms and grep for
> ===
> <gnc.translog>
> ===
> 
> My guess is that will be found to top and tail the delay.
> 
> a quick wc -l on the two log files will tell us (and you) if the same stuff (more or less) was being logged
> 
> then you can look at the time stamp at the top and bottom of each sample and let us know what the difference is.

You can do it yourself, since he posted the two comparable logfiles on Dropbox.
The linux one has 21912 instances of 'xaccTransWriteLog' and the time difference between the first and last it 7 seconds. The Windows one has 21800 instances and the time difference is 17 minutes 1 second.

> 
> If
> 
> * the file was the same at the start of both tests
> * the gnc versions were same major (2.6 seems sensible)
> 
> then there is something else we need to look at.
> 
> Aside:
> 
> How many log lines do you, the reader, have for each
> * hh:mm:ss  INFO <gnc.translog> [xaccOpenLog] Attempt to open disabled transaction log
> 
> here it varies between 10 and 6 with an eyeball check ... but I haven't been careful, it may be 11 and 5.

It alternates between 3 and 5, with the extra two lines being variations on
* 12:20:03 INFO [xaccTransSetDateInternal] addr=07D98158 set date to 1322103081.000000000 Wed Nov 23 21:51:21 2011

He'd already shown you this in your discussion on the Portable Apps forum.

Regards,
John Ralls