I am curious about the current state of handling SSH-Repos under windows using libgit2sharp. Is it possible to e.g. clone a Repo using SSH (git#github.com:libgit2/libgit2sharp.git) ? Thanks
As of v0.21, it's not possible yet.
A Pull request (cf. #852) is in progress to make this happen. Feel free to subscribe to it would you like to be notified of its progress.
The plan behind the scene would be to rely on libssh2 to make this happen.
Related
I'm looking for a way to force a client computer to verify with our MDM filewave server preferably inside a script. Right now the only non gui way I have to force a verify on the client end, is by killing the fwcld service. This works well when ran directly on the client machine, but does not work when ran as a fileset from the filewave server.
The reason I need this, is because Kernel extension policies hang at "Handled via MDM" and will not activate until the next verification. This is way too slow when software is dependent on the kernel extension to install.
If anyone has a solution, either a script, or some other way to make the policy install quicker. I'd really appreciate. I've been banging my head against the wall for too long, the wall is looking pretty rough.
According to FileWave's documentation MDM-managed clients auto-verify with the FileWave server, based on the "tickle interval".
“waits for a specified amount of time (Tickle Interval … ) before it tries to connect again.”
By default this value is set to 120sec. The tickle interval can be changed by using the "SuperPrefs Editor". When using the SuperPrefs Editor only modify the values that you want to change.
Use these instructions to help you upload and apply the new FWpreferences to the (necessary) FileWave clients.
Link to Knowledgebase
**FileWave has moved the Knowledgebase. All links above have been updated. 9/21/22 **
I'm a university student. Currently, I’m doing my Final Year Project that requires a hook that executes automatically on the server side of the repository, before the objects been pulled to the client side, and after the objects have been pushed to the server side, which is "post-receive" hook.
I have done lots of research regard on the hook and I came to know grack and rjgit_grack. Links will be attached below.
Grack : https://github.com/grackorg/grack
Rjgit_grack : https://github.com/grackorg/rjgit_grack
Grack and rjgit_grack is a gem from https://rubygems.org/. Grack project aims to replace the builtin git-http-backend CGI handler distributed with C Git with a Rack application and rjgit_grack is an alternative adapter of grack that support extra hooks for git that have the hook that I needed for my project, "preUpload" hook that is executed immediately before an upload-operation is performed, i.e. before data is sent to the client. But I was unable to get it to work for my project due to various reasons.
Why Git doesn't have a hook that executed immediately before the data is sent to the client? Any advice on getting this hook or any similar function of the hook?
The idea behind Git hooks is that they do at least one of two things: change the state of the repository or the current operation, or optionally block a behavior from occurring.
A hook that executes before objects are pushed wouldn't be able to do either one of these: it has no data about how to change the repository, since it doesn't have any objects, and without any data about what's going on, there's no way that it can usefully determine whether to block a push.
For your project, it might be useful to wrap the git HTTP backend script with a wrapper that executes before the push service starts, or one that looks for the flush pkt-line delimiter (0000) by intercepting the data. You could also patch Git to have the hook you need for your project at the appropriate location.
I'm trying to implement a message sending on my platform.sh server. The idea would be to use the existing hooks:deploy: hook in the configuration file to make a bash curl call that would send a message to an api "Deploy completed".
I use a curl snippet that works like a charm on my local machine, though the message is never sent from the server on a new deploy.
I know that the correct interpreter is bash, since the existing example in platform.sh docs is using bash code. Though, it's indicated in the docs that the disk is in read-only mode during this hook, would that be the cause of the issue?
Thanks a lot!
~SPJ
For posterity: also note that this may already exist by default in Platform.sh you can use the webhook integration to get notified of any activity.
$ platform integration:add --type=webhook --url=A-URL-THAT-CAN-RECEIVE-THE-POSTED-JSON
Details on https://docs.platform.sh/administration/integrations/webhooks.html
Hmm, I thought it was irrelevant, but French text with accents in the API parameters was causing the issue...
Hope this will help others.
~SPJ
Right now I have a Java application running WMI querying event logs, this is painfully slow due to the nature of WMI (ok, not painfully, but it may not be able to keep up with our domain controller), and dllhost.exe gets hammered serving up WMI requests and remote DCOM objects, pretty unnecessary just to read logs.
The next bit of exploration is in Windows RPC calls, but I'm confused... is the Windows RPC implementation (sorta kinda not wonderfully documented for event logs) just another name for WMI? Or will I be receiving raw event log information?
Other than these two methods, does anyone know of any other ways to hook into the event log creation event so I can have the servers automatically push their logs to me? It would be nice if it was something that could be fairly easily implemented in Linux, but I can tamper with WINE and Mono if I have to...
Or would it probably be best to write and deploy scripts on all the servers and have them push it to my program on the Linux box (though now I have to worry about the uptime of all those scripts)?
Or better yet... should I just write a Java service that can plug into the event logs natively and install those on the various windows machines and have it hand off the logs to my central Linux box that way?
Jarapac looks promoising, I'm going to dig a little bit at this and see if Windows RPC performance is up to par and how hard it is to implement. If it's fairly straight forward: Yay! Windows RPC on Linux.
If not, it's simply not possible without your own implementation. :(
I've got a service that acts as a watchdog for several apps/servers. There are no user sessions on this machine. I'd like the watchdog to be capable of beeping on the internal speaker should something go wrong (that'd be my queue to go fix whatever it's complaining about)
when I try the Beep() API on Windows nothing happens - I suspect the problem is that the services session isnt permitted to make noises?
can I make this work? any other ideas for how to make the service alert me?
-CG
Call CreateFile on \device\beep, then send down IOCTL_BEEP_SET (see http://www.koders.com/c/fidFEC3527B9D951559D62722A9C0C603863106CA9B.aspx for details)
It may work if you allow it to interact with the desktop (an option configurable somewhere, I can't remember where).
But personally, I'd have it email me.
Though maybe you could have it use the task scheduling API to schedule a task for yourself, so next time you log on you can see it.
I don't know; you've got a few options. I'd avoid beeping though.
Try sending beep char "\a" to console. Not sure if it will work.
Beeping doesn't seem like a good idea - it might end up driving everyone mad....
I'd also agree about the "interact with desktop" option and you set this in the services parameters see A Windows Service without a template
I'd recommend creating a simple client application that polls that server to query for any problems and returns a set of status messages. Then an appropriate UI would be raised (e.g. balloon on the tray), an email sent, etc. containing any warning or failure messages.
This way you also know that the watchdog itself is running and has network connectivity - if the watchdog dies and/or machine locks up you wouldn't otherwise know.
It also avoids being thrown out of a window when the machine starts beeping continuously just after you go to lunch. [+1 to #mikej] :-)
The poll period should be around half (see Nyquist sampling rate) your minimum required response time.