Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Greg Pringle

Pages: 1 [2] 3 4 ... 9
16
Applications / Re: OpenSSH configuration problem
« on: June 05, 2018, 03:30:03 pm »
The moduli file made the error go away.

With that the the sshd_config file entry:
Subsystem       sftp    C:/usr/bin/sftp-server.exe

makes sftp function.

That allows sftp for the complete file system but I can find little to control the usage.
Then I tried:
Subsystem       sftp    internal-sftp
Match user greg
        X11Forwarding no
        AllowTcpForwarding no
        ChrootDirectory /home
        ForceCommand internal-sftp

but it does not want to work. If it worked then users could be bound to subdirectories for usage.
Here is the server error:
=====
subsystem request for sftp
debug1: subsystem: exec() internal-sftp
'ebug1: Forced command (config) 'internal-sftp
debug1: Received SIGCHLD.
=====


17
Applications / Re: OpenSSH configuration problem
« on: June 04, 2018, 03:43:51 pm »
When I connect to OpenSSH 6.6p1-4.6 on Linux the sftp works.

I wonder if the version compiled for OS/2 is the problem.

18
Applications / OpenSSH configuration problem
« on: June 03, 2018, 01:48:46 pm »
Using Alex Taylors build 5.3p1 and the latest from ArcaOS the OpenSSH will work for clients but has problems with SFtp.

Here is the log from sshd:
=====
debug1: Server will not fork when running in debugging mode.
Connection from 71.246.225.60 port 50909
debug1: Client protocol version 2.0; client software version JSCH-0.1.53
debug1: no match: JSCH-0.1.53
debug1: Enabling compatibility mode for protocol 2.0
debug1: Local version string SSH-2.0-OpenSSH_5.3
debug1: list_hostkey_types: ssh-rsa,ssh-dss
debug1: SSH2_MSG_KEXINIT sent
debug1: SSH2_MSG_KEXINIT received
debug1: kex: client->server aes128-ctr hmac-md5 none
debug1: kex: server->client aes128-ctr hmac-md5 none
debug1: SSH2_MSG_KEX_DH_GEX_REQUEST received
debug1: File /etc/ssh/moduli -> C:/etc/ssh/moduli == /etc/ == UNIXROOT == ETC

debug1: File /etc/ssh/primes -> C:/etc/ssh/primes == /etc/ == UNIXROOT == ETC

WARNING: /etc/ssh/moduli does not exist, using fixed modulus
debug1: SSH2_MSG_KEX_DH_GEX_GROUP sent
debug1: expecting SSH2_MSG_KEX_DH_GEX_INIT
Received disconnect from 71.246.225.60: 3: java.security.InvalidAlgorithmParamet
erException: Prime size must be multiple of 64, and can only range from 512 to 1
024 (inclusive)
debug1: do_cleanup
=====

Here is the log from an SFtp client:
=====
SFTP getList begin
getSession: root 71.246.225.60 22
good session
before connect
getList exc:com.jcraft.jsch.JSchException: Session.connect: java.security.Invali
dAlgorithmParameterException: Prime size must be multiple of 64, and can only ra
nge from 512 to 1024 (inclusive)
com.jcraft.jsch.JSchException: Session.connect: java.security.InvalidAlgorithmPa
rameterException: Prime size must be multiple of 64, and can only range from 512
 to 1024 (inclusive)
        at com.jcraft.jsch.Session.connect(Session.java:558)
        at com.jcraft.jsch.Session.connect(Session.java:183)
        at eforward.util.bin.SFtpClient.getList(SFtpClient.java)
        at eforward.util.bin.SFtpClient.main(SFtpClient.java)
SFTP getList done
=====

Note there is a problem with the prime size.

Does anyone have any insight?


19
Cat5 is good enough for 100m and 100Base-TX and even more. But 100m is the standard and Cat5 is specified for that. For 1000Base-T Cat5e is specified and 100m have to work with it.

Though the 5 and 5e standard differ only in some additional measurement and most of the time the cable is/ was exactly the same. There is absolutely no reason to promote Cat6x or even higher if you do not use 10Gbit or more.

I do extensive networking. Cat6 is needed if the wires are in plenum (ceilings) and pass by interference from other electrical wires. Cat5(e) is not as well shielded and will cause some NICs to start to fail. Otherwise Cat5e will work for 1000m.

I just ran a speed test using 25mb fiber on my ArcaOS side and 5mb fiber on a Linux side.
These crossed the internet for Verizon FIOS to Sprint Business internet. (The Sprint side has an OC48 fiber connection at the Linux side building).

From the Linux side I used ftp and did a get:
150 Opening Binary mode data connection for /efs098va04.tar (20951040 bytes).
226 Binary transfer complete.                                               
20951040 bytes received in 00:42 (479.40 KB/s)

Fromthe Linux side I did an ftp put (To an FTP server I wrote in Java):
mput efs098va04.tar [anpqy?]? y                                                 
229 Extended Passive Mode Entered (|||65102|)                                   
150 Opening Binary mode data connection for /tmp/efs098va04.tar.               
100% |*************************************| 20460 KB  605.22 KB/s    00:00 ETA
226 Binary transfer complete.                                                   
20951040 bytes sent in 00:33 (602.53 KB/s)                                     

Note the speed seems to be limited only by the Sprint speed.

I then tested locally between Windows 10 and ArcaOS between two local subnets using a Netgear router in the middle.
I sent a file from the Windows 10 machine which was not in passive mode:
200 Type set to I.
ftp> put seim_winvmf.exe
200 PORT command successful.
150 Opening Binary mode data connection for /tmp/seim_winvmf.exe.
226 Binary transfer complete.
ftp: 45924170 bytes sent in 6.18Seconds 7425.09Kbytes/sec.

Looks like the speed is fine.                                                                     
                               

20
Running OS/2 is only eCs and ArcaOS. In the box, 2.1, 3.0, Warp Connect, Warp 4, WSEB and maybe a few more. Up until a few fix packs of Warp Connect it was easy to take down an OS/2 machine from the internet by attacking the TCP/IP stack. As of Warp 4 those holes were closed. I have run internet servers with OS/2 and eCS on the internet without any firewalls without any successful attacks. I did have to disable the NetBEUI because it was crackable and I was being attacked. Since I don't actually need NetBEUI it did not matter. The currently running versions are both with and without firewalls.

As to the Cloud. It is cost effective for some servers if a company does not have access to local high speed internet. One problem is that many of the Cloud companies only do Windows or Linux. Also if you have more than a few servers the cost is prohibitive. The setup can be complicated it you are not running normal servers such as a web site. I have customers running at Amazon and have had to implement "double-reverse-relays" to other internet servers to allow connectivity.

21
Setup & Installation / Re: RAM disk and temp folders...what else?
« on: May 14, 2018, 09:53:37 pm »
I  am not sure what happens if you manually delete the files but you could have used the "HiDisk" utility used to create the ram disk in the first place. When using the utility "un-check" the create button and then "check" the drive you want removed and then press the "OK" button.
Then reboot and the ram disk will be gone.

22
Setup & Installation / ArcaOS installer iso
« on: May 14, 2018, 05:57:28 pm »
Does anyone know if the ArcaOS installer is up to date with the patches when a new ISO is generated?

I have an ISO from 5.0.2 Commercial which is dated 2-11-18 and and ISO for a new seat dated 5-14-18 and they are different sizes. The first was 949,945,027 and the second is 949,962,625
Both are "5.0.2"

23
I have installed an Asus HZ270M using a Startech PCIe USB 2.0 card and a X800 video card.
The Startech has a Via chip set so it works fine. It has Sata drives at 2TB for a few drives.

I also installed a Gigabyte GA-Z270-HD3 using a Startech PCI (it had one PCI slot) and a video card.
The problem with the Gigabyte is it only wants to recognize 1/2 TB of the Sata drive.
The Startech PCI is also a Via chip set. This install was the hardest.

24
Setup & Installation / Re: RAM disk and temp folders...what else?
« on: May 09, 2018, 06:10:59 pm »
I use Java extensively for applications.
One is a mail filter which I wrote that buffers the mail. To keep memory usage down it uses a virtual memory vector (FileVect) which I also wrote. This works fine but is much slower than real memory. It has the advantage of being able to buffer huge amounts of data. I recently redirected the virtual vector to use a 3.6 GB ram disk. It is working quite well. I have not checked the speed against a non-virtual vector but it seems close enough.

25
Internet / Re: Firefox - 45.9.0 for OS/2 GA1.1...anyone...anyone???
« on: April 20, 2018, 10:59:55 pm »
0042_01.TRP

Dies on XUL

All "required" elements are loaded.

This note is being written with 45.9.0.2

Seems the developers must be old IBM guys. They would write great programs with terrible installers.

26
Applications / Re: Postresql backup
« on: February 15, 2018, 05:24:06 pm »
Also you might check the logs for errors. You might not like what you see.
There are some known issues. If I get time I will be looking at correcting a few.

The method of back up / restore I mention has the benefit of correctly restoring some characters that can be broken by the other method.

27
Storage / Re: Low level format error with JFS
« on: February 10, 2018, 07:28:48 pm »
My experience is the 38000 files is not a problem unless:

1. the directory is accessed from the desktop.

2. the file system has been written to over 90% over and over and over.

From the command line the numbers can easily be over 100000 files in a directory without a problem.

I use java to read and write directories with huge numbers of files.
The trick is I only use my own utilities for doing directory listings.
These are smart utilities that limit what is actually put into memory.

Also if I need to archive files and the huge directories are actually just archived files I have a utility that.
I use a utility that makes sub-directories by day or hour and move the files to those directories.
Then it will delete files and directories of a number of days old. That number can be say 360 days.
This practice keeps any directory from getting overly huge so listings can be run easily.

28
Applications / Re: Postresql backup
« on: February 07, 2018, 05:40:56 pm »
If the end result is to save the data then there is another solution.

for each table needed do this from the command line:
psql -c "copy eforward.ccbox to '/out/ccbox'" user pwd;

replace user with the database owner and pwd with the password.

fix the path as needed.

To restore do this. Recreate an empty database using the method used to create the original.
Then

psql -c "copy eforward.ccbox from '/out/ccbox'" user pwd;

This is fast and reliable.

29
Hardware / Re: Choosing a Graphics Card
« on: February 02, 2018, 03:49:36 pm »
I have not used this card but it is listed in the Snap readme.

Half height pci-e card available used:

ATI FireMV 2200 PCI-E 128MB PCIE


30
Hardware / Re: Choosing a Graphics Card
« on: January 30, 2018, 04:56:27 pm »
There is an inexpensive X800 card on Amazon for under $40 that works with SNAP and has wide screen.

The card below does not have a fan but instead a huge heat sink (which does get warm).
It needs two slots due to the size of the heat sink.

Look for this:
ATI Radeon X800 Pro PCI-E Video Card (VGA + DVI + S-Video)

Pages: 1 [2] 3 4 ... 9