Hi all, SAMBA pre2.0.7 ./configure --with-quotas Linux 2.0.36 The old disk quota bug is still here. If user tries to write the file to SAMBA share and disk quota is exceeded, SAMBA fills remaining space within the target file with spaces, and writes it to the share. This is "xcopy" behavior. If the program writes small chunks of file, the result may differ. The user is under impression the file was written successfully. While the file is heavily corrupted! Please, sort this out! Best, Sergei.
smakarov@cemi.ims.net.ru wrote:> If user tries to write the file > to SAMBA share and disk quota is exceeded, SAMBA fills remaining space > within the target file with spaces, and writes it to the share. This > is "xcopy" behavior. If the program writes small chunks of file, the > result may differ. The user is under impression the file was written > successfully. While the file is heavily corrupted!Jeremy Allison replied: | It isn't Samba that is doing this, it is the client. [...] Samba *is* returning the over quota error to the client. | What do you execpt us to do here? Is the error returned the same one as the user would get if the initial size-testing write would return (the code implies this, but it occurs at quite a diffferent time...) Similarly, is this the same as they would get if they overflowed their local disk in the same way? (I assume yes) Finally, are the contents of the file the same as they'd get if they overflowed their local disk in the same way? Mr Makarov, I don't understand the "fills remaining space within the target file with spaces" comment.... Surely it's filling it with nulls??? --dave -- David Collier-Brown, | Always do right. This will gratify some people 185 Ellerslie Ave., | and astonish the rest. -- Mark Twain Willowdale, Ontario | //www.oreilly.com/catalog/samba/author.html Work: (905) 415-2849 Home: (416) 223-8968 Email: davecb@canada.sun.com
I give up. I'm going to run NT instead of Samba. It all started out as a nice idea. Save a school a few thousand dollars they'd pay Microsoft and put Linux/Samba up instead of a new NT server. Well, I've tried, I've read your bugs, I've done as much as I can.... and I'm going to give in and run NT. I can run an install program on 30 computers off my shared Samba drives so it isn't like I didn't get very far. I can do multiple drives, have many programs run. But the persistent bugs that applications display on Samba drives are just too much hassle to support. Perhaps there is a winning solution and a way to set it up, but the documentation on the samba site is just lame. There is no solid example of the best way to use the parameters together to form a solid system. The email system on the web returns responses not to use the email system on the web. The newsgroup here didn't even give me on response to my question. The other people out there seem to be getting similar but different errors. I wish it was a workable solution. I wish I wasn't going to throw thousands more dollars at Microsoft. I wish I could tell other schools that Samba worked. But it doesn't, at least not without a lot of support, poking, prodding, and hand holding. The latest error is that students who are using a program they loaded off the shared drive suddenly can't see the shared drive during opening a file and they get an error. But a minute later the drive is there for them again. Along with this error, the server logs PAM_pwdb authentication error for that student. But... the student was using the system already and was logged in. Maybe its not even samba for THIS error. But Samba 2.0.3 started thousands of processes which killed the server. Samba 2.0.6 fixed this situation and runs fine for some programs but not for others. I've tried op_locks=yes (and no). Just thought I'd let you know a place where samba was a failure so in the future you can think about better ways to reach the people who want to use it. Tomorrow I go and install NT.
The attached message contains the beginning of a correction for the following: An NT client sets a file on the server to a specific length, hoping thereby to make sure that there is enough space for the whole file. Unix creates a "holey" file, occupying no additional disk blocks. The NT starts sending blocks The disk runs out of space or quota. The resulting file appears corrupt from the NT side. The code attached tests to see if, at the time the set length is done, that there is enough disk/quota to store the whole file. This is not claimed to be a correction: it is a test patch that may lead us to an acceptable **workaround**. Can the victims of the problem please try this, and arrange to send me logs? --dave -- David Collier-Brown, | Always do right. This will gratify some people 185 Ellerslie Ave., | and astonish the rest. -- Mark Twain Willowdale, Ontario | //www.oreilly.com/catalog/samba/author.html Work: (905) 415-2849 Home: (416) 223-8968 Email: davecb@canada.sun.com -------------- next part -------------- An embedded message was scrubbed... From: David Collier-Brown <davecb@canada.sun.com> Subject: test code for NT overflows... Date: Sun, 13 Feb 2000 17:46:27 -0500 Size: 2906 Url: http://lists.samba.org/archive/samba/attachments/20000214/64cc4936/attachment.eml