Network problem with windows

[Reese Tora]

Hey, guys!  I've got a problem with my computer that I can't figure out.

I can't seem to access web pgaes that require submitting forms using the post method.

What happens is, when I try to submit a form that uses POST(logins, forum posts, etc...) the browser will start to load the page.  With IE the browser ends up becoming unresponsive and I end up having to kill the program.  With FireFox, what happens is the progress bar fails to advance, and the page I submit from will just sit there.  I can cancel loading te page from FireFox, but if I leave it long enough (a minute or so) the FireFox process will jump to using 100kB-160kB of system memory and 48-52% of processor time (the processor is dual core, so I figure it would use 100% on a single core system)

I've tried scanning with AdAware in case it's some spyware messing with me, and Avast is currently running a boot time scan for any viruses that could be causing it.  The problem manifested Thursday morning, and a system restore to Monday did not fix it either

Posting this from my laptop, which pretty much proves that the problem is only with my computer, not with the network connection.

[llearch n'n'daCorna]

That sounds very much like the problem is a proxy of some sort.

Not sure what you'd do about it, though.

[Reese Tora]

well, the story has an OK ending.

After trying a few more fix attempts, safe mode scans, boot scans, etc...

I put in the windows install disk and used the system recovery option (the third one that replaces any windows files with the originals from the CD)  No dice

So I backed up the registry, program files, and documents and settings, formatted C: and reinstalled XP.

The problem is now gone, I have no idea what could ahve been causing it, but it's gone now.

Casualties:

The software for monitoring my UPS had a DLL in the system32 folder (a folder which I purposefully did not back up) the install disk for that finally stopped working (shoulda coppied that floppy! or atleast cehcked it before nuking) and the supposedly same software from the manufacturer's website can't see my UPS.  I will probably go searching for a copy of the older program online some time.

My installations of software I use for web development are gone, but that's not a problem.

I havn't tried out most of my game software yet, but most of that should be intact between program files, documents and settings, and the registry... despite the fact that windows managed to randomize all the drives besides C: (but I moved them back to the proper letter designations afterwards.)

[llearch n'n'daCorna]

The software for monitoring my UPS had a DLL in the system32 folder (a folder which I purposefully did not back up) the install disk for that finally stopped working (shoulda coppied that floppy! or atleast cehcked it before nuking) and the supposedly same software from the manufacturer's website can't see my UPS.  I will probably go searching for a copy of the older program online some time.

My installations of software I use for web development are gone, but that's not a problem.

You mean you don't have a folder with the install files for everything you install on the machine? Drivers, games, programs, utilities, you name it - everything put on the machine.

I havn't tried out most of my game software yet, but most of that should be intact between program files, documents and settings, and the registry... despite the fact that windows managed to randomize all the drives besides C: (but I moved them back to the proper letter designations afterwards.)

The order isn't random. It's complex, but it's not random. First, the primary partition on the primary master disk, then primary partition on the primary slave, then secondary master, then secondary slave. Then all the logical partitions on the pprimary master disk, then slave, then secondary master and slave.

Under Linux, you can have 4 primary partitions, but Windows/DOS only puts one primary in before making an extended partition, then makes all the others logical inside that. Of course, one of the 4 is the extended; and there's a maximum of 16 possible partitions per disk.And this is all under the old partitioning scheme; the newer thing they brought in with Win2k (whose name I can't recall off the top of my head) may have other restrictions I'm not aware of, but MS don't usually tell people about them, other than buried somewhere in TechNet...

So, uh, yeah. Not random. But not exactly simple...

[Tapewolf]

there's a maximum of 16 possible partitions per disk.

I never knew that.

And this is all under the old partitioning scheme; the newer thing they brought in with Win2k (whose name I can't recall off the top of my head) may have other restrictions I'm not aware of, but MS don't usually tell people about them, other than buried somewhere in TechNet...

GPT - GUID Partition Table

There's also BSD slices IIRC, but it tends to make Windows cry.

And yes, an install directory is a really good idea.

[hapless]

there's a maximum of 16 possible partitions per disk.

There might be some limitation that Windows imposes... but the logical partitioning sub-scheme stores headers in something not entirely unlike a linked list, and if there's a limit to the number of chain elements, it's definately larger than 16. My disk goes up to /dev/sda22 now, and i still have couple thousand cylinders free for later allocation.

BSD slices are a concept generally similar to x86 partition table, that originates from architectures that didn't bother with such things on the firmware level and left partitioning to the OS. x86 BSDs took the easiest route of lying to their FS subsystem about the "drive" they're working on, limiting it to one x86 partition with BSD's own partitions inside.

... is GPT working without EFI firmware?

//h(ave a nice day)

[Reese Tora]

You mean you don't have a folder with the install files for everything you install on the machine? Drivers, games, programs, utilities, you name it - everything put on the machine.

I have, actually, but a few files are missing from it... namely, the files that are on disks.  that was the only one on a floppy, though.  Any other software taht is on disk is on a manufacturer supplied CD (and, thus, not subject to the same rate of decay as a burned disk)

To be honest, I'm surprised that my UPS has survived as long as it has with full functionality, as I bought it, what... 6-7 years ago? maybe 8? a while, anyway. (and I never did send in the rebate form. )

I actually did think I had a backup of the files from the disk on my computer, but I can't seem to find it, and suspect I never did.

[llearch n'n'daCorna]

there's a maximum of 16 possible partitions per disk.

There might be some limitation that Windows imposes... but the logical partitioning sub-scheme stores headers in something not entirely unlike a linked list, and if there's a limit to the number of chain elements, it's definately larger than 16. My disk goes up to /dev/sda22 now, and i still have couple thousand cylinders free for later allocation.

I was speaking specifically from memory of the limitations that Windows imposes. According to man fdisk, on Linux, IDE disks are limited to 63 partitions; SCSI to 15. I may have been thinking of that.

It also goes on to say that SUN/BSD disklabel is limited to 8 slices, the third of which should be a "whole disk" slice, and recommends avoiding starting a partition at cylinder 0, because that will destroy the disklabel; this suggests that there are "safety" features of the wintel disk partitioning schema that aren't there in other places; OTOH, you can overlap slices if you so desire. What happens to the contents of the slices is indeterminate, and depends on what you do with the things, but if you want to stand on your own feet, Sun will let you... IRIX/SGI is limited to 16 partitions, two of which are pre-defined, so that's even more weird...

The mention of /usr/src/linux/Documentation/devices.txt as a place for further information doesn't help, as I didn't have that on my system... However, perusal of it doesn't actually shed any more light on the subject.

Go figure. I'm not sure how you managed 22 SCSI partitions, though, as all I can find says that's broken. Hrm...

[Tapewolf]

I was speaking specifically from memory of the limitations that Windows imposes. According to man fdisk, on Linux, IDE disks are limited to 63 partitions; SCSI to 15. I may have been thinking of that.

Those might be limitations within the driver layer.  The command transport shouldn't really care (or know) about the partitioning scheme since it's basically returning a raw addressable space.

[hapless]

SATA drives in Linux are working over the SCSI pass-thru interface. Thus /dev/sd*.
NetBSD/i386 slice table has 16 entries, two of which are reserved. 'c' for whole NetBSD partition, 'd' for whole physical drive. Not sure, but FreeBSD's might be limited to 8.
Also, those 63/15 limits are more probably BIOS/firmware based than anything else. After all, I could take the drive from this pc, plug it into an alpha, and format it for VMS' use with no partitions at all.

[Tapewolf]

I was speaking specifically from memory of the limitations that Windows imposes. According to man fdisk, on Linux, IDE disks are limited to 63 partitions; SCSI to 15. I may have been thinking of that.

...wait.  Where does it say this?  Mine isn't saying that.  What date is yours?

[hapless]

Hmm... On my debian the English version of the manpage doesn't mention such thing, while the localized version says exactly what llerach quoted. Both are dated for 11 June 1998, but there's a fat chance that somebody forgot to update the timestamp while changing contents.

EDIT: Removed in 2008, in this commit.
//h (Yes, I have a lot of time and nothing better to do. )

[Tapewolf]

EDIT: Removed in 2008, in this commit.
//h (Yes, I have a lot of time to waste.)

Sweet.

[llearch n'n'daCorna]

The restrictions were part of the /dev setup. Major and minor node tags, y'know...

Documentation/devices.txt hasn't been updated in 2.6.26; perhaps later?

[Tapewolf]

The restrictions were part of the /dev setup. Major and minor node tags, y'know...

Is that still the case now that we have udev?

[llearch n'n'daCorna]

The restrictions were part of the /dev setup. Major and minor node tags, y'know...

Is that still the case now that we have udev?

/dev/sda and /dev/sdb on my system are using block major tags of 8 and minor tags of 0 and 16, respectively, which is in line with what is listed for SCSI disks in Documentation/devices.txt - you'd have to ask hapless what he has on his machine, but mine appears to still be following the rules as laid out.

Of course, I'm using debian 5.0.2; I've no idea which he's using, nor whether he's using udev or not; my understanding, however, was that udev was to manage device names, not device tags, with a secondary purpose of limiting the excess inodes chewed up by thousands of unneeded devices; it seems to work fine at that, since there's a whole lot of devices that used to be created in /dev by MAKEDEV back on 1.2.13 that aren't there now on here.

Incidentally, you can find the devices list at http://www.lanana.org/docs/device-list/ - it seems at least superficially identical to the version I have here...

[hapless]

Ouch.
Honestly? I looked at fdisk's output, not into /dev/ itself. And it happily "faked" device names that can't exist.
14-22 are NTFS partitions, so I hadn't noticed they're unreachable from linux yet.
Ah, well, maybe it'll teach me to check twice... or maybe not. Anyway, my first mistake on this forum was after the fifth post I wrote... that's a new personal record.

Regards,
//h

[llearch n'n'daCorna]

Ah. That explains a lot.

Yeah, fdisk will happily make device names that don't exist. I'm not sure quite how that works out in the Real World, though. ;-]


Glad we all figured out what was going where, though...

[hapless]

Whew. 14 days, straight on the border of decency... though, after reading some of the RP threads, I understand the reason behind having the hard limit set at full month

So, seems that 2.6.30 finally worked around the limit:

(...)
brw-rw---- 1 root disk   8, 14 2009-08-05  /dev/sda14
brw-rw---- 1 root disk   8, 15 2009-08-05  /dev/sda15
brw-rw---- 1 root disk 259,  0 2009-08-05  /dev/sda16
brw-rw---- 1 root disk 259,  1 2009-08-05  /dev/sda17
(...)

259 block   Block Extended Major
        Used dynamically to hold additional partition minor
        numbers and allow large numbers of partitions per device

//h, still feeling stupid for not checking before the previous post.