Sniffing the local traffic when only the Direct Connect client was trying to establish a connection, resulted in connections over port 4111 (TCP) only.
So I guess the problems of not being able to connect from outside is caused by the dumb SpeedTouch Wireless router.
Okay, I’ve literally spent HOURS trying to get the port forwarding to work with the new router (Belkin N1) and nothing! First of all, the WLAN IP doesn’t seem to be correct to begin with, (one time it was a myspace.com server?!) the hostname ‘Adam’ is not recognized by the router for some odd reason and there’s no way to assign static IPs from the router that I’m aware of. There are no firmware updates btw.
I’m going to email Belkin about this and if they don’t offer a solution, I think the best thing to do is make the server become a DHCP server as well :/
See if the thing has a firmware update available. Usually such crap comes with the default firmware, but they fix it later on.
Let’s hope the router will be able to work as DHCP relay client so that those who connect wirelessly can obtain the IP from Adam (if we had to go that way).
By the way, it’s been about 4 days that the Internet traffic at home is very slow, when I have Azureus running (downloading at 10kB/s, uploading at 17 kB/s) — and my connection is 512kbps.
I hope QualityNet didn’t start throttling P2P traffic … that would be very bad for them. They wouldn’t be living up to their name…
Where’d the join in post go? I’m using ShakesPeer on the Mac it is working wonderfully. To connect though, have to enter the host as HOSTNAME:PORT for it to work.
I’m working to get a client running on my pc at home to share off the NAS.
When I got up this morning, I tried to browse your files, Verlihub kicked me off the server I think and sent me to a public hub lol. I couldn’t get back on, I had to close the client and reconnect. I was using (DC++) linuxdcpp. It worked fine after that.
Here’s a question.. Does DC++ count as P2P traffic?
So there wasn’t a problem with transferring files? Did you get the file list in one piece?
Last Thursday, we installed a new 1TB disk, partitioned it, and got it ready to enter the array, so that we can swap out 2x 500GB disks.
Knowing our luck, it was inevitable to go through hell. For some reason, the BIOS didn’t like the new disk, and decided that the disk order should be changed, too.
3baid figured out that the BIOS changed the order of disks, which is why we couldn’t boot up for over an hour. Unfortunately, we faced much more problems.
When the system booted, the new disk partitions were added as spares to both arrays, and on the first array, we marked one 500GB disk as faulty to remove it later. The recovery process proceeded as it should.
Later that day, 3baid tells me that all disks are marked as spares on the first array, and that it’s no longer active!!!
We tried to bring it up, but it refused. Also, the disk names changed again, and a disk disappeared from the arrays.
Today, we were able to bring up the arrays again and the data is intact (since we disabled Samba to not allow anyone to write to the disks).
We still have one disk missing. It could be either the disk’s fault, or some problem with the SATA port or cable.
Either way, to prevent future problems with disk naming, we’re cooking a plan to assign static names to disks by using their UUIDs.
In the first group, you can see that they all have the same UUID. This is nor a mistake. This is the UUID of the array they belong to, and this is what helped us keeping the data intact, when the device names changed.
In the 2nd run, you can see that there are 2 different UUIDs. This is because these 2 devices are from 2 different arrays.
So, how to get the UUID of the device itself, and not the array it belongs to?
I have no clue. Yet.
But I noticed that the UUID of the device is not the same as the superblock UUID, but all devices of the same array have the same… Check this:
mdadm -E /dev/sda1 | more
/dev/sda1:
Magic : a92b4efc
Version : 00.90.00
UUID : 46d09607:abeee337:d5d632b1:97538b27
For other partitions of the same array, the UUID is the same, but it’s different than the one reported by “blkid” o_O
I’d love to hear an explanation, although I’ll be more interested in finding a solution to get the UUID of each partition (assuming there is one).
From the Goolgizing I’ve been doing, I noticed that SCSI disks do have UUIDs, but there’s no mention of SATA. Maybe they don’t have any?
If that’s the case, then the whole idea is down the drains…
2 days ago, after 5 days of pain, we were able to fix the disk issues. One cable was always loose…
I told 3baid to use super-glue, but he didn’t take me seriously. I think I’ll super-glue the bloody cable to the port when 3baid is not in the dewaniya…
Data corruption is not a friendly thought…
As for assigning static names to disks, one of the guys in #linux (DALNet) suggested using /dev/disk/by-serial which identifies each hard disk by its vendor, model and serial number, and further more, detects each partition, which is exactly what we need.
I’ll poke around this weekend and see how to do it with udev.
We were having problems stopping the RAID arrays after un-mounting the logical drive. Always got an error that the device is busy or being used.
I suspected that the Volume Manager is the one hogging the arrays, and I was right. I looked around on how to stop the volume group manager, and the solution was:
vgchange -a n
-a is for volume group availability, and “n” is for no. To re-activate, use “y”
I haven’t tested it, since the guys are abusing Adam, but I will tomorrow morning when they are snoozing in their mothers’ basement. I mean their beds.
May 16, 2008 at 12:13 am |
K,
Putting details for each router brand is just too much :/
We can say things like set up a port forward rule, for both TCP & UDP. Then people should refer to their manuals on how to achieve that.
May 16, 2008 at 4:36 am |
MBH, I confirmed that the hub can’t be accessed remotely and I think it’s because of the router :/
I can access regular hubs, but Adam gives me timeouts.
May 17, 2008 at 9:44 am |
It could be requiring other ports than the one we setup.
Next time I go to the dewaniya, I’ll sniff the traffic when connecting to the hub and see which ports are being used.
May 25, 2008 at 7:46 pm |
Sniffing the local traffic when only the Direct Connect client was trying to establish a connection, resulted in connections over port 4111 (TCP) only.
So I guess the problems of not being able to connect from outside is caused by the dumb SpeedTouch Wireless router.
June 13, 2008 at 3:26 am |
Okay, I’ve literally spent HOURS trying to get the port forwarding to work with the new router (Belkin N1) and nothing! First of all, the WLAN IP doesn’t seem to be correct to begin with, (one time it was a myspace.com server?!) the hostname ‘Adam’ is not recognized by the router for some odd reason and there’s no way to assign static IPs from the router that I’m aware of. There are no firmware updates btw.
I’m going to email Belkin about this and if they don’t offer a solution, I think the best thing to do is make the server become a DHCP server as well :/
June 13, 2008 at 11:02 pm |
See if the thing has a firmware update available. Usually such crap comes with the default firmware, but they fix it later on.
Let’s hope the router will be able to work as DHCP relay client so that those who connect wirelessly can obtain the IP from Adam (if we had to go that way).
By the way, it’s been about 4 days that the Internet traffic at home is very slow, when I have Azureus running (downloading at 10kB/s, uploading at 17 kB/s) — and my connection is 512kbps.
I hope QualityNet didn’t start throttling P2P traffic … that would be very bad for them. They wouldn’t be living up to their name…
June 13, 2008 at 11:56 pm |
There’s no firmware update. The connection’s been fine on our side.
June 17, 2008 at 5:07 pm |
IT WORKS! The Firmware updates WORKS! :D
The tech support sent us a beta firmware via email :]
We’re beta testers now :P
June 23, 2008 at 10:31 am |
Where’d the join in post go? I’m using ShakesPeer on the Mac it is working wonderfully. To connect though, have to enter the host as HOSTNAME:PORT for it to work.
I’m working to get a client running on my pc at home to share off the NAS.
Awesome work guys.
June 23, 2008 at 12:07 pm |
The DC Hub is on the main page.
June 24, 2008 at 6:56 am |
When I got up this morning, I tried to browse your files, Verlihub kicked me off the server I think and sent me to a public hub lol. I couldn’t get back on, I had to close the client and reconnect. I was using (DC++) linuxdcpp. It worked fine after that.
Here’s a question.. Does DC++ count as P2P traffic?
So there wasn’t a problem with transferring files? Did you get the file list in one piece?
June 24, 2008 at 3:23 pm |
Yes, everyone’s reporting the redirection and the same thing happens to me when I need to close the client and open it again.
Yes, DC++ is considered as P2P traffic unfortunately :/
Yes, I was able to get the entire file from you.
August 17, 2008 at 4:22 pm |
Last Thursday, we installed a new 1TB disk, partitioned it, and got it ready to enter the array, so that we can swap out 2x 500GB disks.
Knowing our luck, it was inevitable to go through hell. For some reason, the BIOS didn’t like the new disk, and decided that the disk order should be changed, too.
3baid figured out that the BIOS changed the order of disks, which is why we couldn’t boot up for over an hour. Unfortunately, we faced much more problems.
When the system booted, the new disk partitions were added as spares to both arrays, and on the first array, we marked one 500GB disk as faulty to remove it later. The recovery process proceeded as it should.
Later that day, 3baid tells me that all disks are marked as spares on the first array, and that it’s no longer active!!!
We tried to bring it up, but it refused. Also, the disk names changed again, and a disk disappeared from the arrays.
Today, we were able to bring up the arrays again and the data is intact (since we disabled Samba to not allow anyone to write to the disks).
We still have one disk missing. It could be either the disk’s fault, or some problem with the SATA port or cable.
Either way, to prevent future problems with disk naming, we’re cooking a plan to assign static names to disks by using their UUIDs.
Details will come up later.
August 17, 2008 at 4:52 pm |
The program “blkid” can display the UUID of a disk/partition/device.
user@Adam:~$ blkid /dev/sd[abcef]1
/dev/sda1: UUID=”0796d046-37e3-eeab-b132-d6d5278b5397″ TYPE=”mdraid”
/dev/sdb1: UUID=”0796d046-37e3-eeab-b132-d6d5278b5397″ TYPE=”mdraid”
/dev/sdc1: UUID=”0796d046-37e3-eeab-b132-d6d5278b5397″ TYPE=”mdraid”
/dev/sde1: UUID=”0796d046-37e3-eeab-b132-d6d5278b5397″ TYPE=”mdraid”
/dev/sdf1: UUID=”0796d046-37e3-eeab-b132-d6d5278b5397″ TYPE=”mdraid”
user@Adam:~$ blkid /dev/sd[ag]1
/dev/sda1: UUID=”0796d046-37e3-eeab-b132-d6d5278b5397″ TYPE=”mdraid”
/dev/sdg1: UUID=”98d7c538-1f5e-5131-e164-0bd68cf09905″ TYPE=”mdraid”
In the first group, you can see that they all have the same UUID. This is nor a mistake. This is the UUID of the array they belong to, and this is what helped us keeping the data intact, when the device names changed.
In the 2nd run, you can see that there are 2 different UUIDs. This is because these 2 devices are from 2 different arrays.
So, how to get the UUID of the device itself, and not the array it belongs to?
I have no clue. Yet.
But I noticed that the UUID of the device is not the same as the superblock UUID, but all devices of the same array have the same… Check this:
mdadm -E /dev/sda1 | more
/dev/sda1:
Magic : a92b4efc
Version : 00.90.00
UUID : 46d09607:abeee337:d5d632b1:97538b27
For other partitions of the same array, the UUID is the same, but it’s different than the one reported by “blkid” o_O
I’d love to hear an explanation, although I’ll be more interested in finding a solution to get the UUID of each partition (assuming there is one).
From the Goolgizing I’ve been doing, I noticed that SCSI disks do have UUIDs, but there’s no mention of SATA. Maybe they don’t have any?
If that’s the case, then the whole idea is down the drains…
August 20, 2008 at 4:21 pm |
2 days ago, after 5 days of pain, we were able to fix the disk issues. One cable was always loose…
I told 3baid to use super-glue, but he didn’t take me seriously. I think I’ll super-glue the bloody cable to the port when 3baid is not in the dewaniya…
Data corruption is not a friendly thought…
As for assigning static names to disks, one of the guys in #linux (DALNet) suggested using /dev/disk/by-serial which identifies each hard disk by its vendor, model and serial number, and further more, detects each partition, which is exactly what we need.
I’ll poke around this weekend and see how to do it with udev.
We were having problems stopping the RAID arrays after un-mounting the logical drive. Always got an error that the device is busy or being used.
I suspected that the Volume Manager is the one hogging the arrays, and I was right. I looked around on how to stop the volume group manager, and the solution was:
vgchange -a n
-a is for volume group availability, and “n” is for no. To re-activate, use “y”
I haven’t tested it, since the guys are abusing Adam, but I will tomorrow morning when they are snoozing in their mothers’ basement. I mean their beds.