I use ipython notebook on the browser of my laptop that connects to a remote ipython notebook server in my office. The problem is that if the network is down for a very short period, the connection will be closed. Then to reconnect to the server, I "just" need to refresh the browser, but which is actually quite annoying if the size of the notebook is large.
The question is can I increase the connection timeout (just like such options for ssh-server), so that the connection is not lost at all even if the network is down for a short time.
You might want to check your jupyter_notebook_config.py and check the following variables in the file.
cull_idle_timeout
shutdown_no_activity_timeout
You can checkout the documentation for these variables here
They shouldn't be enabled by default, but since you are connected to an external server your connection is likely dropped until its refreshed. There might be security reasons for this. Might be a good idea to speak to your Sys. Admin as well before you enable them :)
Related
guys, I need your help. Here is the problem I face now:
I have a Dell XPS laptop and install windows 10 and Linux Mint 18 on it. To keep the clock consistent with my time zone, I have used the following command in Linux:
timedatectl set-local-rtc 1 --adjust-system-clock
The problem now is, in windows 10, I can only connect to Gooogle or gmail during the beginning of the reboot. After long time, or after a suspension of windows 10, I can not connect to Google service anymore. It will report "can not estabish private connection". I can connect to other website normally. The same problem also exist for my dropbox. It will report "can not establish a secured connection".
I have checked the certificate of https for Google services, it seems that the browser locates to a wrong certificate when the connection is failed.
The problem disappear if I use a VPN connection.
I guess the problem is related to clock setting in my windows system, but I don't know how to solve it.
any suggestion is appreciated.
Solved.
The problem is that I use manual DNS settings, so the time in my computer and the server is not consistent. Just need to convert to auto DNS.
Sorry for the interrupt.
I am forced to use a visual-svn-server that is located in our windows domain. The problem is that it is super slow to use with windows client. Weird thing is that the same repository is very fast with linux client. The difference is like 3sec vs 90sec. I know somebody should fix the server, rather than me trying to fix the client, but i have no change of doing that.
So, to debug the problem I did some package capture with wireshark and it seems like windows, when doing 'svn up' (on up to date repository) does quite much ldap-negotiations before actually talking again with the actual svn-server. This takes time. Linux svn client when doing 'svn up' is not doing any ldap calls. The problem is not on my machine, but on all my colleagues windows clients too.
I tried forcing the svn client to 'basic' auth with configuration option http-auth-types (http://svnbook.red-bean.com/en/1.7/svn.advanced.confarea.html), but it didnt help. I figured that would be basic, no ldap, http-basic-auth. I am able to confirm that the setting is included, since setting it to 'digest' says that authentication method is not available. But even that takes about 60s, so my guess is that its doing the ldap-wacko stuff before trying to do the authentication.
The subversion client i am using is 1.8 serie from tortoise svn official build. I did try also slicksvn client and it did have the same problem. The svn versions shows ra_serf is handling the https requests and my repository is visual-svn server located at https://my_server_intra_dns_name/
When opening the address with browser, its again fast as it should, so problem should not be with dns or similar.
I am linux guy, so i am bit lost with windows, but does anybody have an idea wtf is going on here?
---- edit ----
I had also linux as guest operating system on the windows host, and inside that linux doing svn up was about 3s, compare that to native windows 'svn.exe up' that took over minute !
If a Windows machine has a limited connectivity to the Internet, then you may notice the delay when running Subversion client command's against a remote repository over HTTPS.
Using a traffic analyzer you can notice, that the delay happens when Windows attempts to access ctldl.windowsupdate.com and gets a timeout. Windows attempts to access ctldl.windowsupdate.com to check Certificate Trust List (i.e. Certificate Revocation List). With limited Internet connectivity, Windows may be unable to access it thus resulting in these delays.
If it's not your case, then I suggest contacting VisualSVN's support team for investigation.
In my case it was due ot Windows proxy settings - that you set in IE (I use TortoiseSVN client, and Visual SVN Server was set to use basic authentication).
After I've set up IE proxy settings accordinlgy (automatic for me, but for you it might be something different) initial delay was gone.
It helped even though the svn server is on local LAN and I have checked with Wireshark if the traffic goes over proxy. In Tortoise I have proxy disabled. Why it helped with my issue - no idea.
The initial delay I had was 11-13 seconds. Now next to none.
And I am not using ssh client.
Go to http(s) location of your SVN server using your browsers: IE, Fireofx, whatever, and if the response is quick then it is very possible that is an svn client problem, or due to some similar settings (similar to your browser settings).
For instance IE was slow (IE was set up for local connection only previously), Firefox (with proper proxy settings) was OK - and SVN server IS local (sounds like some sort of network/firewall/routing issue to me, but proxy settings helped me).
I'm using SSH tunneling to connect from a Windows 7 machine to a remote Postgres database.
The tunnel works fine when I use pgAdmin to connect to the remote database. However, I haven't managed to set up an ODBC DSN to connect to the remote database.
I'm creating a System DSN with the same server address/username/password/port/ssl mode as pgAdmin but when I click "Test" I get: "Could not connect to server; No connection could be made because the target machine actively refused it."
Any ideas on where to start debugging? I've ruled out Windows Firewall and remote server configuration, since I can connect with the exact same information in pgAdmin.
Edit
The problem resolved itself after reinstalling Windows.
When debugging this the first thing to do is to start testing the PostgreSQL server with the psql command line utility. This avoids issues of drivers etc to see what is actually going on. Generally speaking my troubleshooting process is:
Can I ping the server?
Can I connect with psql?
When I test my connection via the control panel, does it work?
Now, assuming the problem is on #2, the first thing I would do is try disabling firewalls on both sides (they can be brought up and tuned after trying with both down). This rules out misbehaving firewalls and if it shows this to be a problem then you have the answer on where to troubleshoot.
In my experience Windows firewall can cause all manner of problems. I have even seen it block web pages halfway through the web page. I don't think you can rule out Windows Firewall based on the information you have provided.
If this happens again, the first things to do include disabling both any antivirus software and firewalls.
The other possibility would be a bad ODBC config (maybe port 5423 instead of 5432) or a badly installed DLL of some sort. The latter may be fixed (or not) using sfc /scannow or reinstalling the odbc driver, or other actions.
This is my situation:
I have a linux server/media center with a windows client.
My goal is to remote control rhythmbox amongst other things.
I've done this using plink (windows based cli ssh toy).
The problem is that starting up an ssh session logging in and sending a command is understandably slow as hell. When I had a windows server I used a tool called psexec which was almost instantaneous.
Is there any way to speed this process up? Either somehow sending the commands with the login request which should show some improvement. Or by maintaining a persistent ssh connection which I can use. (plink dcs at the end of the command).
More info: On my windows machine I'm using a bat like:
plink -ssh -l username -pw pass myipaddress "/home/username/bin/skip"
On my linux machine the skip bash file is something like:
//needed to get around a x11 error caused by controlling rhythmbox over sshif its an ssh connection copy the dbusaddressfirhythmbox-client --next //the cli wrapper for rhythmbox
Further Research:
The only way to go seems to keep an ssh connection open/maintained as a service. This seems doable as there is a demand due to setting up ssh tunnels (to bypass firewalls). From there I'd need a way to send the command line commands to this existing connection or reuse that connection.
The other option is of course to NOT use ssh. Hell I already have a connection through samba file shares and there is no lag there. I bet I could put a service linux side that checks for a modified file. Then have an ap client side that modifies said file. Amazingly hacky but so far it seems like the best option. And by best I mean the only one that cuts control lag. There has got to be a better way than this, I can't be the only nerd using linux as a media-center that wants remote controls. This kind of moves the topic from stackoverflow to superuser but that's ok.
You could user an SSL certificate to get rid of the login part. Alternatively, build yourself a small HTTP server which uses an "exotic" port for controlling your media player (amarok, btw, has one build-in)
Switching to something like mpd will bypass the ssh issue, although I give no guarantee that changing tracks will be any faster.
If anyone is curious, I ended up implementing an http based server with php to execute commands server side. And client side I used curl.exe to allow me to have nice click-able buttons without the overhead of a web-browser.
Also nice since it allowed me to implement an in browser UI which is great to use from any machine with internet, ones that don't have ssh installed. And works wonderfully from my phone as a remote control (which I can use from a country away if I so chose...)
I know it is not strictly a programmer question, on the other hand, I would really like to be able to do a simple svn up on the production servers, which would save us a lot of hassle.
Or production servers are within a corporate network, and access is only allowed through a gateway server. From that we can access the actual servers via Remote Desktop. Because of this maintenance unfriendly setup we use a sftp server to get our files to the production server (where the ftp server alwys resets last modified date), so it is complicated to determine which files have changed.
So if it would be possible to setup a ssh tunnel from the actual server through the citrix connection, it would ease the whole update process.
(On further option which I consider is to use bazaar to push our updates to the sftp server and then pull the updates from there, but unfortunately there is no working svn plugin for bazaar available [on windows])
Okay, not wanting this to remain unsolved, I posted the following on the usenet in alt.os.citrix:
G'day everyone
Someone asked me this question. I haven't a clue. Any ideas?
"the connection I have is PC -> Citrix Access Gateway -> actual Server. My question is, if it is possible to establish a ssh connection from my PC to the actual Server."
The response I got from a Catherine Jefferson was:
Should be. I do this all the time using an Access Gateway VPN, then ssh to servers inside the company LAN.
What version of the Access Gateway is this user running? I work for Citrix, might be able to get a more specific answer for him or her.
So may I suggest you hook into alt.os.citrix (probably via Google Groups if you don't have an NNTP tool) and buy into the conversation.
--Bruce
I do this a lot. The tunnel is like:
ssh -nvCXAN -L 1494:10.0.2.39:1494 www.starshipping.com &
Then, you need an ICA file that points to localhost.
[WFClient]
Version=2
[ApplicationServers]
Connection To Citrix Server=
[Connection To Citrix Server]
WinStationDriver=ICA 3.0
TransportDriver=TCP/IP
ClientAudio=Off
DesiredColor=8
Username=yourusername
Domain=yourdomain
Password=xxxxxxxxx ; has to be encrypted. I pulled from another ICA file generated internally
ScreenPercent=97
TWIMode=OFF
Address=localhost
This all worked very fine on OSX until Citrix gave out a new version of the Citrix client which is now called the Citrix Online Plugin. Haven't been able to get it to work since then. Meanwhile, our VPN has been working fine, so haven't worried too much about it.
Citrix over SSH, How to Create a Key Pair for SSH Authentication Using PuTTYgen, Terminals. It seems that the answers are "out there" but not readily found. That was just five minutes in Google. HTH.