Unable To Complete A Track Scan With Alpha5

cmx · Post by **cmx** » Tue Jun 19, 2007 11:03 am

Running P4, 1GB ram, XP with mix of M4A and MP3.
I have 121969 files as reported with MM2 and itunes.
No other programs were running during these attemps.

Initially I got hang-ups with debug emails generated and sent. Evidently the db generated in previous alpha was corrupt in some way. Was never able to successfully start alpha 5.

Deleted database and ini files.

A5 then started succesfully.

Upon the initial Scan with a fresh A5- I left it running overnight...the next morning A5 had closed itself.

Upon re-opening, A5 reported about 106000 files had been scanned.

Upon rescanning, A5 reached the previous number of previously scanned files very quickly, about 6-7 minutes...BUT when it appeared to be starting to scan previously "unscanned" files, A5 slowed to approx 1 files every 5 secs or more...it scanned several files in this manner, then A5 simply "vanished" no visible error was generated by MM itself, nor was anything apparent generated by windows.

Upon subsequent tries at rescanning A5 seems to "vanish" immediately, just as it hits the previously unscanned files.

This behavior upon rescan was consistent...it appears to possibly scan a few "new" ones on each subsequent try...but then quickly "vanishes".

The particular files that are being scanned when the problem seems to occur were scanned in earlier alphas(although I have never gotten a complete, working, error free scan with MM3) and were scanned successfully in MM2 and itunes.

I then tried skipping the specific directory folder being scanned when the "crash" was occuring(a folder with many, many subfolders of compilations)....this did not seem to help...the "vanish" still occured at about the same number of total files scanned...its hard for me to determine because I have to be looking at it when it happens....there is no warning before A5 "vanishes"....

I can't really determine how to proceed from here without guidance.

Please advise. Thanks for any help.

(An aside...it's curious, I believe I was able to get a complete scan with the initial alpha or alpha2...there were subsequent database issues of course, like the way things were grouped etc...but I WAS able to get a complete scan of all tracks...after the intital successes in a1 and a2 I have not been able to get a successful scan of my files...)

rovingcowboy · Post by **rovingcowboy** » Tue Jun 19, 2007 11:16 am

sounds like the page file is getting filled and crashing monkey.
that use to happen to me on the win98 when i was getting too many scanned it just up and locked or crashed.

i made the swap file bigger and got more files scanned but not all of them. so i had to just scan so many folders at one time.

i chose half of what i did when it crashed. and using only half of them fixed the crashing.

so make your page file larger and see if that helps.

either monkey wants the larger page file or the programs monkey kicks out of the ram want a larger page file. but i would try the page file first let us know.

Teknojnky · Post by **Teknojnky** » Tue Jun 19, 2007 11:17 am

Dev's will need a debug log to best track down the problem(s).

I've sent several logs from alpha 4 (where I also had add/rescan issues).

I am in the processs of rescanning with alpha 5 a new from scratch database and have ran into a couple of the dreaded debug error windows.

Eventually I will get around to sending a debug log of my alpha 5 scan.

cmx · Post by **cmx** » Tue Jun 19, 2007 11:47 am

tekno...the autodebug system they have in place does not seem to be operating with this specific problem....as I described A5 simply vanishes NO debug errors generated....

cowboy...thanks for the suggestion....

I set up system monitor to watch the pagefile %...and yes seconds before A5 vanished the page file access maxed out....

I have my pagefile set to "custom":
initial 1536
max 3072

This is three times available ram, I set it here in the pqst following MSFT recommendation....I'm am simply not knowledgeable enough to make decisions or assumptions concerning this stuff...

I will now try "system managed"....and see how that works.

Please advise further on this matter. Is this something the devs can deal with?...

rovingcowboy · Post by **rovingcowboy** » Tue Jun 19, 2007 12:10 pm

change that page file to have one less number then the max. does.

make it min 3071 and max 3072

see if that helps.
since windows will only have one mb to flex with.

when the min is sat that low windows clears all the page file mb's set above the min. and use's it for normal hard drive memory.
when the need for it comes up.
windows will reset the extra mb's as needed.
making it for ever rasing and lowering the used page file. so it might be having trouble with the resetting of it when needed.

also check the same panel area for how the quantums are set up.?

if they are set to make your programs work better. then switch them to make the system to work better that will allow the page file to use more cpu cycles since its part of the system.

of course rebooting is needed after all these settings.

Teknojnky · Post by **Teknojnky** » Tue Jun 19, 2007 12:17 pm

cmx wrote:tekno...the autodebug system they have in place does not seem to be operating with this specific problem....as I described A5 simply vanishes NO debug errors generated....

I understand that the debug dialog did not get presented, however you can and should still generate a debug log as per the sticky post in the bug forum.

The debug dialog that pops up, may provide snapshot information, but it does not show the whole picture of what happened leading up to that point. And if MM disappears well there not even info from the debug dialog.

The only way for the dev's to see the whole picture is a debug log.

cmx · Post by **cmx** » Tue Jun 19, 2007 12:25 pm

tek, I might be confused aboput this, but I tried the debug process described in the old post you mentioned and got errors......when I asked about the debug overlay described in the post, I was told it was actually included in the alpha....

either way...I need clarification and help to proceed.

cmx · Post by **cmx** » Tue Jun 19, 2007 12:29 pm

roving....I think it might have been wishful thinking....

Upon maximizing the pagefile to 4096 AND letting the system deal with it....same exact crash behavior by A5...

What are the affects of such a high pagefile size?

This computer is used for work and some of the programs are resource intensive...don't want to set the overall focus to system over programs...

also note that it was "pages/sec" that spiked...the actual pagefile usage % and Usage peak really didn't seem to move or have issues at all throughout....

I will now try doing the files in smaller segments....

Any further clarification, ideas or help greatly appreciated....

rovingcowboy · Post by **rovingcowboy** » Tue Jun 19, 2007 12:32 pm

did you set the intail page file amount up like i said?

that amount you posted is what i am running on my win98 system windows xp needs more intail setting whic is just the min amount. that should help what monkey seems to be doing is hogging the banana's again. the devlopers will have to slim him down again. make him run on less memory again like they did last year.

Teknojnky · Post by **Teknojnky** » Tue Jun 19, 2007 12:36 pm

I have not noticed any spike in memory usage nor have I had MM 3a5 disappear on me.

The first thing I would suspect related to spiking memory would be problems reading a specific file or files.

Narrowing it down to a specific folder would be ideal, but I do understand that is not very convenient espeically when your not sure which out of possibly thousands it could be.

edit:

btw, the debug viewer does work fine with mm3 alpha including alpha 5.

I am using debugview version 4.64 now available from microsoft (they bought out sysinternals) @ http://www.microsoft.com/technet/sysint ... Suite.mspx

monkey hi fi · Post by **monkey hi fi** » Tue Jun 19, 2007 12:48 pm

sorry cmx i was typing that last one when you posted the other message.

what are the effects of more page file space. ? better running programs and system.

but the quantums will only improve one or the other a little more then the page file.

example is in a 3d game you are to have a shiny lake on the screen, but you only have the color blue. with out the shine.
give the page file more space, and you get the shine on the 3d game. some programs will send you an alert if there is not enough swap file or page file, for them to use.

just the page sectors were in heavy use? makes me think of you needing extra stacks added. the only way i know how to adjust the stacks is by use of the xteq program in its tweaking stuff it has an option to adjust the stacks. i set my win98 up to 20 stacks then up to 24 stacks i am not sure where i stopped at but it was under 30 spare stacks.

roving cowboy / keith hall.

cmx · Post by **cmx** » Tue Jun 19, 2007 12:49 pm

roving...same behavior occurs no matter tha pagefile size setting...currently set to 4095mg

tek: yeah there seems to be no memory spike, that was my error in reading the system monitor.

I will download and install the link you posted...BUT what are the "relevant log files"...could MM staff give some specific instruction here....hate to just fly blind and wast time.

Feels like it is specific file/files...but i'll be damned if it still crashes if I skip whole folders....arrrrgh

Teknojnky · Post by **Teknojnky** » Tue Jun 19, 2007 1:05 pm

in order to get the relevant debug logs, you must run debug viewer before starting MM, make sure 'capture' is on (use the capture menu > capture events, or the magnifying glass toolbar buton), start the debug MM (either alpha or the debug ver of 2.5x whichever your trying to generate the log for), let mediamonky startup and try to recreate the error/problem.

After the problem is reproduced or mm crashes/freezes, you ca save the log and send it to the devs.

Depending on log size, you may need to zip it up.

Steegy · Post by **Steegy** » Tue Jun 19, 2007 2:10 pm

@cmx: your page file is waaaaaaaaaaaaaaaaay too big (yes RovingCowboy, that's my opinion). It's best to set it back so that Windows takes care of it. That way, if Windows notices it runs out of memory, it will dynamically increase the page file size. The "recommended" setting often said is 1.5 times your physical memory. So it you have 512 MB of RAM, the page file size should be around 768 MB. Too much usage of the page file (the part of the memory that gets written to the hard drive) is actually sometimes worse than before (because the hard drive is much slower than RAM). Therefore some servers tend to use less page file size, sometimes even 0. Anyway, if you're using a big page file (and even if you're not), it's important that your hard drive is enough defragmented, so it runs smooth enough.
The most important thing is too look at Task Manager to see how the CPU and RAM usage are. If they're too high (certainly the memory: the memory usage and VM size), then errors are known to occur (in any programs).

cmx · Post by **cmx** » Tue Jun 19, 2007 2:46 pm

yeah steegy,,,pagefile not the issue....making some final comparisons and tests...but it look like a couple of M4P files are the culprits! Thanks to tekno for getting me hooked up with the debug thing...stay tuned

OK yes....2 days and 6 hours later.....

Isolated to two folders.
Both contain M4P files. (Jimmy Buffet is one of course...lol)

If I deselect these two folders I get a complete scan of my files.

Note:
Several other folders of M4P files WERE scanned in ok. YET, if I try to play one of the scanned M4P files alpha5 generates an error email(sent in).

One other interesting note...

alpha5 returns 121238 total files (artist node)
MM2.5 returns 121969 total files (artist node)

not sure what one is reading the other isn't and vice-versa.....

I have dbqview.log's of the cold crashes on the M4P's if the devs want it. I can also send example of files if needed. I don't want to send all this stuff unsolicited. Please advise.

Thanks tekno....now I can start picking things apart..I can already see there are still issues with Compilations/Various artists

Unable To Complete A Track Scan With Alpha5

Unable To Complete A Track Scan With Alpha5

M4P Scan Problems