blistovmhz

I dig the zero-conf in principle, but if it doesn't work...

80 posts in this topic

1 hour ago, Nick Bolton said:

If all goes to plan, there should be a new update available to download tomorrow morning.

That sounds great! I hope they go to further stabilize the Ubuntu client.

Share this post


Link to post
Share on other sites
3 hours ago, blistovmhz said:

2.x is currently deciding that the lowest cost path between my two machines which are directly connected to each other, is through a VPN with 120ms latency. I had to perform logical host isolation on all clients to convince 2.x to stop doing this.

I think the symptom here is mouse lag. How many other users are struggling with zeroconf taking the wrong route?

We have a solution on our roadmap, so depending on how many people this affects, we will prioritise this.

So, how many people are seeing a laggy mouse cursor in S2? (where they didn’t see lag in S1)

Share this post


Link to post
Share on other sites
49 minutes ago, Nick Bolton said:

I think the symptom here is mouse lag. How many other users are struggling with zeroconf taking the wrong route?

We have a solution on our roadmap, so depending on how many people this affects, we will prioritise this.

So, how many people are seeing a laggy mouse cursor in S2? (where they didn’t see lag in S1)

I see severe lag on wireless connected clients, despite sub 10ms latency. Seems hit/miss and I haven't confirmed that this isn't just another "traversing the VPN" issue.
My biggest concern is the zeroconf pooping itself every time a client disconnects, requiring all clients to be restarted.

Share this post


Link to post
Share on other sites
13 hours ago, Nick Bolton said:

 

@Steve Williams Anything to add?

 

Only that we do take all the feedback seriously, and that we plan to regularly release updates with bug fixes in over the next 6 months,

Share this post


Link to post
Share on other sites

@blistovmhz Could you give 2.0.5 a try? It was released today. We haven't added a settings screen yet (coming soon), but it's more stable. Let me know what you think!

Share this post


Link to post
Share on other sites
34 minutes ago, Nick Bolton said:

@blistovmhz Could you give 2.0.5 a try? It was released today. We haven't added a settings screen yet (coming soon), but it's more stable. Let me know what you think!

No dice, but I figured out part of the issue pretty definitively.

I've sent a log here: https://synergy-logs.symless.com/4c0e0e5a10c43ceec337731b9cfcf629/logs/29302-2018-02-08T11-10-37.log

For context, I have three clients connected on the same LAN.

  • octillion
  • the_rain
  • quadrillion

Octillion is always supposed to be master (though it'd be amazing if whichever client was actively mousing, temporarily became master)...
When Octillion is master and the cursor/focus is in Octillion, if I press the power button on Quadrillion, Quadrillion becomes master and it does not suspend as it should. If however, Octillions cursor is focused on Quadrillion, I press Quadrillions power button and Quadrillion suspends and Octillion remains the master.
Same issue with any other laptop or desktop. Any time the power button or lid switch (any ACPI power event as far as I can tell) is thrown on a non-master client while focus is not on that client, the ACPI event is intercepted by master, and master control is given to the machine that should have been commanded to suspend/shutdown.

Share this post


Link to post
Share on other sites

Initial test results between OSX 10.12.6 and Linux Mint (Debian) 18.3 looking good. Cursor lag's decreased and copy-and-paste seems more reliable.

Share this post


Link to post
Share on other sites
13 hours ago, blistovmhz said:

Octillion is always supposed to be master (though it'd be amazing if whichever client was actively mousing, temporarily became master)...
When Octillion is master and the cursor/focus is in Octillion, if I press the power button on Quadrillion, Quadrillion becomes master and it does not suspend as it should. If however, Octillions cursor is focused on Quadrillion, I press Quadrillions power button and Quadrillion suspends and Octillion remains the master.
Same issue with any other laptop or desktop. Any time the power button or lid switch (any ACPI power event as far as I can tell) is thrown on a non-master client while focus is not on that client, the ACPI event is intercepted by master, and master control is given to the machine that should have been commanded to suspend/shutdown

This doesn't sound right. Have you opened a support ticket for this with logging from each of your machines? 

Share this post


Link to post
Share on other sites

No, I gave up because the ticket system wasn't working. I forwarded my log a few comments back.

Share this post


Link to post
Share on other sites
1 hour ago, blistovmhz said:

the ticket system wasn't working

Do you mean you couldn't reply to the email, or that you couldn't submit a new ticket?

If you're having problems with the form, please use: [email protected]

Share this post


Link to post
Share on other sites
On 2/8/2018 at 11:36 AM, Nick Bolton said:

@blistovmhz Could you give 2.0.5 a try? It was released today. We haven't added a settings screen yet (coming soon), but it's more stable. Let me know what you think!

Hi Nick,

You mentioned that a settings screen is coming soon. Can we take that to mean that it's coming before 2.1? I bought synergy just today, and the constant host-swapping (e.g. "local input detected") is making it barely usable. Being able to force one machine to always be the server would resolve the problem. I'm trying to decide if I should stick it out or request a refund / 1.x license.

 

Edited by Fope
typo

Share this post


Link to post
Share on other sites
21 minutes ago, Fope said:

Hi Nick,

You mentioned that a settings screen is coming soon. Can we take that to mean that it's coming before 2.1? I bought synergy just today, and the constant host-swapping (e.g. "local input detected") is making it barely usable. Being able to force one machine to always be the server would resolve the problem. I'm trying to decide if I should stick it out or request a refund / 1.x license.

 

I REALLY like the idea of automatic host swapping, in theory... if/when it works.
But yea, the "active host" detection doesn't work at all. I can't even guess how it's determining who's active. I can wag physical mouse A around all day, but host B still thinks it is the master. Then Host A goes to sleep or connects to a VPN, and now Host C thinks it's master (Host C doesn't have any physical HID inputs... ).

I think the only patterns I've seen is that very often, when a client comes online, it is automatically master for a while.

Share this post


Link to post
Share on other sites
3 hours ago, blistovmhz said:

I REALLY like the idea of automatic host swapping, in theory... if/when it works.
But yea, the "active host" detection doesn't work at all. I can't even guess how it's determining who's active. I can wag physical mouse A around all day, but host B still thinks it is the master. Then Host A goes to sleep or connects to a VPN, and now Host C thinks it's master (Host C doesn't have any physical HID inputs... ).

I think the only patterns I've seen is that very often, when a client comes online, it is automatically master for a while.

In my case the second (preferably 'client') machine is a laptop, which obviously has a keyboard and trackpad that I can't just unplug. However, it sits on a shelf behind my desk and there cannot possibly be any input occurring, unless the raw trackpad input is providing some noise or something. Typically for me, though, the host switch happens as the mouse transitions from one screen to another, so I'm compelled to think that synergy itself is causing cursor movement that is being mis-characterized as local. Without hooking up a debugger, though, I'm just speculating.

Share this post


Link to post
Share on other sites

Okay, I did just find SOMETHING.

Lets say you have devices A,B,C. A is currently the master and we'll refer to it's mouse as mouseA.
Move mouseA to clientC. Now use mouseC to try to navigate to client A or B. Doesn't work.

Instead, move mouseA back to clientA, and now move mouseC to client A or B. Works fine.
Now you're stuck using mouseC though, until you put mouseC back into clientC and then move mouseA.

The automatic host detection just doesn't work if any other client currently has control over the active client.

Now, as to why disconnecting a client results in everything breaking, I still haven't found a pattern.

Share this post


Link to post
Share on other sites

Popping back in to report that I'm also experiencing a second mode of zero-conf failure reported in this read: Disconnecting/rebooting/etc causes synergy 2.0.5 to wind up in an untenable state. My (preferred host) laptop is in a dock, from which it is removed several times a day to attend meetings and the like. When I come back and plug it back in, both the (preferred) server and the second machine seem to think they are the server, and this stalemate continues until one of them backs off by restarting the service. 

 

I've had synergy 2.x for just over 24 hours, and I've run out of fingers on which to count the number of times I've had to manually restart the service. lol come on

Share this post


Link to post
Share on other sites
14 hours ago, blistovmhz said:

Lets say you have devices A,B,C

What OS are you running on each of these devices?

Share this post


Link to post
Share on other sites
10 hours ago, Steve Williams said:

What OS are you running on each of these devices?

A is usually Linux, B usually Windows, and C usually Linux.

Doesn't seem to matter though if I'm just all windows or homogeneous. 

Share this post


Link to post
Share on other sites
On 2/14/2018 at 2:17 PM, Fope said:

Popping back in to report that I'm also experiencing a second mode of zero-conf failure reported in this read: Disconnecting/rebooting/etc causes synergy 2.0.5 to wind up in an untenable state. My (preferred host) laptop is in a dock, from which it is removed several times a day to attend meetings and the like. When I come back and plug it back in, both the (preferred) server and the second machine seem to think they are the server, and this stalemate continues until one of them backs off by restarting the service. 

 

I've had synergy 2.x for just over 24 hours, and I've run out of fingers on which to count the number of times I've had to manually restart the service. lol come on

Yea I found another one as well. Switch any client from wired to wireless and you're done. Synergy won't work again until you reboot and restart all clients.

Share this post


Link to post
Share on other sites

Synergy 2.06 was released Monday. Does it work better for you? So far, with 2 startups and several hours of use, it is working perfectly. I haven't had to open the UI on any of the 4 computers.

Share this post


Link to post
Share on other sites

I requested and received a refund last week, so I can't report on whether the new version fixes anything. I'm still interested in following development, though, so I'm a little dismayed that the change log isn't up yet.

 

The one thing I can report is that compared to 2.0.5, 1.8.8 is perfection.

Share this post


Link to post
Share on other sites

No change with 2.0.6 in a homogeneous environment for me. Last WINDOWS client to connect, always gets master control. Also seems like if I even move a physical mouse/keyboard on any windows box, it gets master. Master never comes back to the linux machine unless I manually force it.
Making me lose my mind. Everyone at the office has just quit using it and are trying to find alternatives. :(

*edit*. Confirmed behaviour. Linux is supposed to be master, but if I CLICK with a physical mouse on any windows client, or any windows client reconnects to the *cough* network, that client always takes master control. This would be GREAT if Linux clients also took master control back automatically, but they do not. I have to fire up the "config UI..." and manually take control again.

 

Edited by blistovmhz

Share this post


Link to post
Share on other sites
5 minutes ago, blistovmhz said:

No change with 2.0.6 in a homogeneous environment for me. Last WINDOWS client to connect, always gets master control. Also seems like if I even move a physical mouse/keyboard on any windows box, it gets master. Master never comes back to the linux machine unless I manually force it.
Making me lose my mind. Everyone at the office has just quit using it and are trying to find alternatives. :(

*edit*. Confirmed behaviour. Linux is supposed to be master, but if I CLICK with a physical mouse on any windows client, or any windows client reconnects to the *cough* network, that client always takes master control. This would be GREAT if Linux clients also took master control back automatically, but they do not. I have to fire up the "config UI..." and manually take control again.

 

Sorry about that. I have this same problem and it drives me mad. I’m pushing for this to be a priority, as it’s crippling the workflow for any Linux running in client mode. Have you raised a support ticket? This helps us to speed up the bug fix. 

https://symless.com/contact/customer-support

Share this post


Link to post
Share on other sites
2 minutes ago, Nick Bolton said:

Sorry about that. I have this same problem and it drives me mad. I’m pushing for this to be a priority, as it’s crippling the workflow for any Linux running in client mode. Have you raised a support ticket? This helps us to speed up the bug fix. 

https://symless.com/contact/customer-support

Also with 2.0.6, if a windows client goes offline and comes back, it does not rejoin the network automatically. If I restart the synergy service on the newly re-joined client, and open it's synergy GUI, it again hijacks control but also crashes the linux synergy clients entirely. The linux synergy service must then be re-started and the gui must be opened. at this point, the linux Gui shows that it can't connect to it's own local service until you manually take control from that client. Then all of a sudden it can connect to itself.

Share this post


Link to post
Share on other sites
On 1/31/2018 at 5:15 AM, Nick Bolton said:

Actually, we're currently focusing entirely on Synergy 2.0 bug fixes. We will soon be focusing on new features for Synergy 2.1, which will include documentation. This is due in about 6 months. Thanks for your patience.

How about you release 1.9.0 until 2.x is ready for use?

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now