UMSDOS maintenance - without losing existing FREESCO

[Island]

Looked around for information on UMSDOS, particularly any maintenance commands, but without much luck. I wanted to check that UMSDOS is ok on my FREESCO, because it seemed to get into difficulty. (I recall a previous thread about UMSDOS commands but I think it was using the previosu forum and searching doesn't find it for me.) Here's my problem:

Running FREESCO 042 and it has been quite reliable; several packages also loaded, including EXIM handling mail for a DynDNS domain. FREESCO sits on a DOS drive which also has a fairly full W98SE installation on it; I estimate there's around 500MB for FREESCO. The logs overflowed, with two symptoms:
(i) On boot, FREESCO reported 'more than 4000 entries in directory', and
(ii) ran very slowly, reachable from inside the LAN, but WAN connections weren't possible (I've seen this effect before when the logs get too big)

Logged in through SSH and used mc to remove most of the historic log files. But, FREESCO WAN connection still not operating (unable to even 'ping'). I'd like to check that the UMSDOS filesystem is undamaged, and I'd like to also check how much free space there is (for FREESCO, not just what free space there might or might not be in the underlying DOS). I don't want to do anything that might damage the FREESCO installation; the box is fairly inaccessible and difficult to reload.

Dingetje's wiki warns that umssync is a dangerous command, so I haven't run that; I haven't been able to find any documentation for it either. There is another command, udosctl, but I couldn't find which of its commands might help me.

Anyone have any experience of checking the integrity of UMSDOS on their FREESCO? I've a vague memory of some 'trick', was it 'deleting' the --Linux- entries and UMSDOS rebuilds them, or something like that?

regards, Island

[strampke]

Go to the dos prompt and type scandisk /? or /help or (I forgot)
Read
Then make a choice like: SCANDISK /all /autofix /nosave /surface /nosummary or whatever and after that your disk will be happy again

Use JV16powertools to delete unneeded files, then delete register garbage, delete a lot of programs
Then do a defrag on your HD and your HD will be even more happy.

[Lightning]

The above is quite accurate. UMSDOS is just plain DOS and it requires the standard DOS commands to make any repairs to it. The UMS part of it is just an extra file in every directory that gives file ownerships and permissions and visibility for files and directories that normal DOS does not have. Which umssync and udosctl both are Linux applications for manipulating those files and that is not what you are wanting to accomplish.

I suspect that "scandisk" at a DOS command prompt with a surface scan along with a "defrag" will resolve the issues, which can even be done from within Windows 9x running. For future en-devours I would strongly recommend going to an ext2 file system and using the ext2 package from a floppy to move the system to a hard drive. However the downside of doing that is that any existing partitions will be erased. But the advantage is a self de-fragmenting file system and built in tools for repair with no reliance on DOS or DOS commands.

[Island]

Stramke, Lightning, thank you both; SCANDISK found and fixed errors in C:\ROUTER. It also found some lost directories, and some file fragments.

FREESCO runs, and can 'ping' from a direct SSH login, so the outbound route is ok. The control panel webserver works. But normal 'operations' don't seem to work, in that FREESCO isn't routing onwards (to the WAN) the packets from computers on the LAN.

I've checked the network tab in the control panel. The default route is correct. (That's partly why the 'ping' worked, of course.)

Port 123 traffic from internet is being accepted (I think this is NTP and I had installed NTP 4.1.1 a few days ago).

The logs are showing most DNS replies (port 53) being rejected (the DNS reply identifying the destination of the 'ping' are being accepted but this request from the FREESCO box itself). I think this is why the LAN machines cannot get web access. I think the LAN machines are requesting DNS resolution, DNS is being forwarded correctly, but the replies are being rejected. So, does this mean the firewall is not working quite correctly? I'm guessing some base data has been corrupted when the filesystem became damaged.

On the network tab of the control panel there's a whole bunch of stuff; reported, it seems, from the command:

ipfwadm -I -lne

I cannot see anything that seems to refer to port 53 - but I don't really know what I'm looking at, there.

Other entries in network tab are that:
all forwarding from the LAN is going to 0.0.0.0 (I think this is ok and is forwarding the LAN computers' DNS requests, for example);
there are no active masquerading entries;
there are no entries in the port-forwarding table (this is correct);.

Anyone have any suggestions about how to get the traffic forwarding/replying back working again? Preferably - really preferably - without re-installing FREESCO?

regards, Island

[Lightning]

From what you are saying the router is probably routing traffic, but no DNS resolution.
The solution to this problem could be one of two things "most likely".
Run the "setup" and go into the server tab. Select DNS server and it should either be set to "s" or "n". If it is not one of those two then the configuration file was corrupted and should be reset and then saved. Once that is completed go to the command prompt and type these three commands.

rc_dns restart
rc_dhcpd restart
rc_masq restart

Then check your Internet connection again. If it still does not work then attaching a "report.txt" in a reply is going to be required.

[Island]

Lightning, thanks for the response. I think I misled you, by mentioning the DNS. The example - which I've logged, below, 'happens' to be an external DNS req from a LAN machine to (Google's) DNS server, but is being rejected. I quote it only as an example of the failure of LAN machines to transact over the WAN. (It happens to be a DNS transaction because that's the initial transaction for most WAN sequences.) Here's the relevant log:

Code: Select all

Sep 15 12:27:32 - dnsmasq[2432]: reply www.google.com is 209.85.148.147
Sep 15 12:27:32 - dnsmasq[2432]: reply www.google.com is 209.85.148.99
Sep 15 12:27:32 - dnsmasq[2432]: reply www.google.com is 209.85.148.103
Sep 15 12:27:32 - dnsmasq[2432]: reply www.google.com is 209.85.148.104
Sep 15 12:27:32 - dnsmasq[2432]: reply www.google.com is 209.85.148.105
Sep 15 12:27:32 - dnsmasq[2432]: reply www.google.com is 209.85.148.106
Sep 15 12:27:32 - dnsmasq[2432]: reply www.l.google.com is 209.85.148.147
Sep 15 12:27:32 - dnsmasq[2432]: reply www.l.google.com is 209.85.148.99
Sep 15 12:27:32 - dnsmasq[2432]: reply www.l.google.com is 209.85.148.103
Sep 15 12:27:32 - dnsmasq[2432]: reply www.l.google.com is 209.85.148.104
Sep 15 12:27:32 - dnsmasq[2432]: reply www.l.google.com is 209.85.148.105
Sep 15 12:27:32 - dnsmasq[2432]: reply www.l.google.com is 209.85.148.106
Sep 15 12:28:07 - kernel: IP fw-in acc eth0 UDP 63.240.161.99:123 192.168.12.70:123 L=76 S=0x00 I=0 F=0x4000 T=40
Sep 15 12:28:08 - kernel: IP fw-in acc eth0 UDP 94.23.207.136:123 192.168.12.70:123 L=76 S=0x00 I=0 F=0x4000 T=52
Sep 15 12:28:08 - kernel: IP fw-in acc eth0 UDP 88.190.17.126:123 192.168.12.70:123 L=76 S=0x00 I=0 F=0x4000 T=38
[ repeats for several lines ]
Sep 15 12:29:23 - kernel: IP fw-in acc eth0 UDP 193.55.167.2:123 192.168.12.70:123 L=76 S=0x00 I=0 F=0x4000 T=42
Sep 15 12:29:25 - kernel: IP fw-in rej eth0 UDP 8.8.8.8:53 192.168.12.70:48712 L=126 S=0x18 I=52797 F=0x0000 T=42
Sep 15 12:29:26 - kernel: IP fw-in rej eth0 UDP 8.8.4.4:53 192.168.12.70:48713 L=126 S=0x18 I=26928 F=0x0000 T=42
Sep 15 12:29:31 - kernel: IP fw-in rej eth0 UDP 8.8.8.8:53 192.168.12.70:48712 L=126 S=0x18 I=34672 F=0x0000 T=42
Sep 15 12:29:32 - kernel: IP fw-in rej eth0 UDP 8.8.4.4:53 192.168.12.70:48713 L=126 S=0x18 I=34673 F=0x0000 T=42



192.168.12.70 is FREESCO's WAN IP (behind an ISP WAN router under our control).

As I see this, following the 'ping' test to http://www.google.com while logged in under SSH - which succeeded - a bunch of NTP transactions (which happen to use dnsmasq to resolve DNS) succeed from the FREESCO box running its NTP-4.1.1, but the attempts (by another machine on the LAN behind FREESCO) to transact with Google's DNS (8.8.8.8 and 8.8.4.4, port 53) fail. My reading is that it's an input from WAN ('fw-in') that is being rejected, so the outbound request 'to' Google's DNS had already succeeded because it had already triggered a response from their DNS server. (The LAN machine making these requests is configured with static IP and to use Google's DNS explicitly.) Incidentally, FREESCO is usually configured with the DNS as 's', not 'e', and I'll check that is still the case but the outbound port 53 req to 8.8.8.8 does suggest, I think, that FREESCO is not intervening on directed DNS requests from the LAN.

I take what you say about the report txt, but before we do that, is there a general command to stop and re-enable the whole LAN->WAN->LAN NAT 'process' (I'm not sure what that's named). I had restarted the Firewall using the control panel, without effect.

Quick question then, before a report.txt, can I re-initialise the NATting system?

regards, Island

[Island]

Lewis, it has started working (I think). At least, some LAN machines can communicate through FREESCO; I can't test them all, yet.

I haven't changed anything functional, though. I used setup to switch on more detailed logging, and I added the 'expert' setup items to the setup menu - one of which was to turn FW/NAT on and off (I hadn't known that was there). Coupled with a couple more reboots, the system seems to be back up. A little bit unsatisfactory, I think, I'd feel a bit more confident if I'd been able to find what was wrong. But still, the most important thing is that the system seems to be is running.

Thanks for your help, as always

regards, Island

[Lightning]

I am glasd the system is working and the commands I gave are a pretty good reset of the most important components. As for a command to re-initialize the entire system. From the command line

rc_main restart