Frequent crashing from "corrupted files"

#46 - Sept. 6, 2012, 7:35 p.m.

[Copied from another thread]

For those who are encountering this issue, here’s what would be super helpful:

Run a memory diagnostic such as MemTest86 (make sure it’s thorough; some memory-errors are fleeting and require sustained use or special conditions).

If the memory diagnostic shows errors, post here saying so! After that, I suggest contacting your manufacturer. = )

If no errors are found, then please open the “Guild Wars 2” folder inside of your Documents folder. You should see one or many files with names similar to “Corrupt-ab48d7ad.dat”. Attach some of those to a post here; if you have many, just pick a few. Also, please attach the file “CorruptArchive.log”, which will be in that same folder.

There’s a good chance that there’ll be too many responses for me to personally examine all of the files, so I apologize in advance if I don’t personally respond to your post. = (

Also, I have no reason to believe this problem would be related to graphics cards/drivers.

#51 - Sept. 6, 2012, 6:06 p.m.

onmuumno.2918:

Seems to me the files size is mismatched inside the client. At the bttom of the document posted it states this several times.

Ah, don’t mind the phrasing in that log-message — it’s misleading. What it really should say is something more like “Decompressing failed after processing 1234 of 5678 expected bytes”. = )

CorruptArchive.log + Corrupt-*.dat contain the information required to see what went wrong.

#53 - Sept. 6, 2012, 6:10 p.m.

fcv.1542:

Do you think it could be a problem with some HDDs or SSDs ? Like I said before, I moved the folder from my Corsair Force GT “SSD” to my 500GB “HDD” and I have not received any error yet. But I was having problems and errors every 15 minutes when I was running the game from my SSD.

Any number of hardware-issues could cause problems like this, including HDDs and SSDs. I’m not familiar with what diagnostic utilities are available for SDDs, but you may want to check your manufacturer’s website and give it a shot. If it turns up any problems, I’d be interested in knowing.

#54 - Sept. 6, 2012, 6:12 p.m.

Spy.2437:

seems that .dat and .zip files aren’t allowed as attachments

Hmm, that’s unfortunate… I’ll talk to some people about what mechanisms we have available for this sort of thing. The forum-technology is outside of my area of expertise.

#55 - Sept. 6, 2012, 6:26 p.m.

Thanks for the files, Spy. At the moment it’s a manual process examining them, so I’m going to put in some effort to automate this. It’ll take some time to gather information on this issue, so I thank you all for your patience! Don’t count on answers tonight. = (

#68 - Sept. 7, 2012, 12:30 p.m.

Spy, I grabbed your files and ran my scripts on them. Screenshots of the comparisons on those three files are attached below. (Note that I assembled these screenshots by hand, so I can’t exactly do this in particular on a regular basis.)

In each of the three blocks from these corrupted files, there was only a single-bit error. The error was in the exact same ‘column’ (address 0x??AE) and always occurred for the bit representing ‘4’ (evidenced by 1 changing to 5 and A changing to E).

Apologies if this is a bit technical, but I wanted to share my exact findings with you. To me this strongly suggests an issue with your RAM; I have no theories otherwise at this time. Believe me when I say I’d love for this to be a bug so I can fix it and get all of you playing for days straight without issue. = (

(Still looking into what options we can use to receive more files.)

#69 - Sept. 7, 2012, 12:36 p.m.

Also, as Remembrance.2639 pointed out in another thread — make sure to use MemTest86+ for 64-bit platforms.

#246 - Oct. 2, 2012, 11:09 a.m.

Hey, all! For anyone who’s still afflicted by this and hasn’t identified hardware as the culprit, I’ve got something I’d like you to do.

Run the game with the command-line parameter of “-analyze” (minus the quotes). Warning: this process is a bit crude on the UI-front; it reuses the text “Verifying archive” and will open a copy of Notepad upon completion (you may want to turn off auto-play temporarily). When Notepad opens, go ahead and copy the entirety of the contents and paste them in a reply to this thread. To keep this forum-thread skimmable, I’ve pre-emptively added spoiler-tags to the output.

Note: I’ve now noticed that if you don’t have any corrupted files, you’ll be presented with the not-so-helpful error of “Download failed!”. On the plus side, this means that there’s no record of corruption!

Keep in mind that this analysis is only meaningful for crashes with a description like “A file was corrupted in the archive”. The results it shows are unfortunately not going to help diagnose any other issues.

With this tool in place, there’s no need to attach/post your CorruptArchive.log, Corrupt-*.dat, or ArenaNet.log files in this thread. The output of the analysis should contain the important information already.

My personal experience has shown that most if not all of these cases stem from hardware-issues, though I recognize that there are a number of you who have run numerous diagnostics showing no results. I don’t have an answer for you, but let’s see what this tool shows!

#249 - Oct. 2, 2012, 2:57 p.m.

and that were just some … i couldnt post whole one since there are so may corruptions
also today it didnt crash at start, some days i cant even enter the game
today i played about 5 mins before the crash (no WvW)
i added the whole thing
hope this helps …

Sorry to hear that you’ve been having so much trouble, eater. = (

Would mind detailing what tests you’ve run and for how long? The analysis shows corruption identical to other cases where we’ve concluded that hardware was at fault.

EDIT: When I say ‘concluded’, I mean that it was demonstrated otherwise that the machine had faulty hardware or bad settings. Immediately after correcting the issue, the corruption-problems disappeared.

#253 - Oct. 2, 2012, 7:41 p.m.

Detahmaio.2014:

I had to attach the -analysis in a document because i exceed 5000 characters.

Detahmaio, have you run any hardware-diagnostics such as memory-tests? If so, which tests have you run and with what settings? Your analysis also shows what I’d expect to see for issues with RAM — whether it’s the memory itself or settings related to it (such as timings).

If you’ve overclocked any settings on your computer, you could consider turning them back down and see whether the problems go away.

#259 - Oct. 3, 2012, 1:37 p.m.

Due to the overlock, you may have file corruption that needs repaired first.

An excellent thing to point out. Thanks, Cindas!

To ensure understanding, I’d like to also point out that you should only ever need to run -repair once after you think you’ve remedied the issue with your system. If you continue to hit archive-corruption crashes and/or prompts to repair your archive, then the issue hasn’t been fully solved yet.

Detahmaio, would you mind sharing the specific diagnostics you’ve run on your machine? Which memory-testers, which set of tests, how long the tests ran for, etc. Even if you’ve already done thorough tests with nothing turning up, the information is useful to me for understanding how the corruption does and doesn’t present itself. = )

Also, it looks like you’ve got well over 200 instances of corrupted files. Would you mind running with -analyzeFull and attaching that result to this thread? The regular -analyze option limits the results to keep the log-size low. If the log-file is too large for attaching to this thread, let me know and I’ll arrange something for you.

#266 - Oct. 4, 2012, 5:25 p.m.

Detahmaio.2014:

Here is the -analyzeFull file

Thanks for posting the full file! Looking over it, I’m seeing two distinct one-bit errors appearing over and over again at the same relative locations. There are also some completely scrambled files, but that is the sort of effect you can see when compressing data that differs even by one bit.

Detahmaio.2014:

I ran the windows memory diagnostic it took about 7-9 minutes to complete after I put everything to stock

Sometimes memory-issues won’t reveal themselves except under heavy stress. If you’re able and willing, I’d suggest something like an overnight run of MemTest86+ with its full suite of tests. Another option is to run a similarly long session of Prime95, which is highly sensitive to memory-errors.

#272 - Oct. 5, 2012, 8:39 p.m.

Bakau.3849:

I ran Prime95 all night and no errors at all – everything passed. There is nothing wrong with my ram.

Hey, Bakau — would you mind running with -analyze and posting the results? I notice that you’ve reported encountering this a lot, so you could also consider running with -analyzeFull and attaching the report rather than posting it inline. This will at least give us an idea of what the corruption looks like and give us some clues.

Do you happen to have GW2 installed to an SSD? If so, you could try moving it to the HDD and see whether it helps. I’ve seen a few posts and heard various anecdotes about some SSDs having issues with “silent corruption”. As of yet I don’t have any solid sources on this, but it’s easy enough to try.

#282 - Oct. 8, 2012, 9:21 p.m.

NaughtyProwler.8653:

I’m stuck in a repair loop… again. The only way I was able to fix this before was to reinstall completely. I had no issues prior to the lastest patch, so I’m going to go and chalk this problem up to the patching process.

I’m afraid I’m not seeing any posts from you in this thread detailing what things you’ve tried other than reinstalling (on the same drive, I think). Here are some things to check out:

- Try moving the installation to your HDD if you have one (and haven’t already tried this)
- Run MemTest86+ overnight with the full suite of tests enabled
- Run with ‘-analyze’ and post/attach the results here

Post here with any findings (or lack thereof)!