Flex Lockup on 1.10.16

Norm - W7CK

Well, I had my first crash this morning.  It appears that the conditions were slightly different than what others have reported.  The Flex 6700 didn't crash, just SmartSDR.

As I reported yesterday, I have been playing around with the configuration of panadapters and slices throughout the day while watching the memory used by SmartSDR steadily increase.  Late yesterday, I took a snapshot of Task Manager with only 4 panadapters open and 4 slices.  Here's the snapshot.



Then sometime this morning while I was gone, the following.



Please note:  I am running SmartSDR v1.9.13.173.  I was able to restart SmartSDR without restarting the 6700.  The 6700 itself did not lock up or emit any noise.  Only SmartSDR failed.

I restarted SmartSDR, memory usage was initially at 124.8MB but continues to climb.  The last time I looked, it was around 1543.8MB.

It looks like a memory leak to me!

I've been running this version for quite a while now without any issues.  The only thing I can think of is that there have been a few Windows updates and it is possible something changed within the OS that is making it hard for SmartSDR to release memory. 

Seems to be an issue though!

NM1W

two flex6700 crashes; One yesterday, one this am. Neither was a radio crash with the annoying tone; both required the long press of the power button to shutdown.
In both cases the 6700 had been on for about 4 hours; It was decoding 4 wsjtx slices;
win10, 1607; i7-6850; 32GB ram; machine was pretty idle.
This is almost as annoying as the dax corruption....

Norm - W7CK

Today everything has been running just fine.  Memory usage is running around 600MB+- and seems to be behaving quite well.   I have no idea why things went crazy yesterday.  I even rebooted and restarted everything yesterday and still had memory usage issues until it finally crashed.  Today, all is well......  This makes troubleshooting very hard!

Norm

NM1W

One other item I noted from both crashes, is that upon restart the number of slices open isnt what it was before the crash, nor on last power down; I run with all 8 panadapters, each with a slice.. On restart I had pans, 6 slices. They seemed a bit random as to which pan had which slice..

Tim - W4TME

We REALLY hate that when it happens.  It makes finding root cause even more challenging.

Norm - W7CK

Tim,

Same here.  I have not been able to duplicate a crash.  If you look back through my posts, I was able to take screen prints of the memory usage and sort of document what I thought was happening.  Now, all seems to be working fine again!

Now when I open 8 slices and make a bunch of band changes the memory usage continues to climb while I'm making the changes.  Once I stop making band changes, some of the memory does seem to get released and usage stabilises.  When I was experiencing the crashes, the memory usage would continue increasing even after I stopped making changes and would eventually crash.

I'm guessing there was some sort of update from MS that conflicted with SSDR for a short period of time and MS must have made another change and problem resolved?  Who knows?  

Like you said, this makes it really challenging finding the root cause!

Norm

Norm - W7CK

Flex staff,

What is considered the normal range of memory usage for SmartSDR connected to a 6700 running 8 slices?

Norm

DaveC

Just an update NO CRASHES since 3/29. BTW Windows 10 32 bit. There has been several Windows update since my last crash. Maybe a Windows issue?

Eric-KE5DTO

Once the waterfalls have had time to get to full storage, it is something like 110MB per Panadapter.  Something around 1GB of memory is typical with 8 Pans running all at once after about 15 min.

We have indeed found and fixed a memory leak associated with changing bands as previously reported here.  This fix will be included in the next release.  That will also contribute to the memory in v1.10.16.

James Skala

Great job Eric.

Norm - W7CK

Thanks for the update Eric.  Hmmmm, now all I have to do is wait...........  Geez, that's hard to do.....

NM1W

Another 6700 lockup;  Had to reboot the radio with a button press; It appeared to only need a short press; Rig had only been on about 10 minutes.. All the pans and slices came back this time.

Tim - W4TME

A normal button press that shuts down the radio indicates that the firmware was still running (the radio had not crashed).

Steve Jones gw0gei

Following on from my last crash (see above - no tone but required long button press in middle of 4m contest) i submitted another ticket. Closed down now as ack as software defect which is logged.

I have closed down the radio and pc for 24 hrs on tuesday this week whilst away on business. Its been on since wednesday evening and did 80m cw contest wednesday night and 80m rtty test on thursday night. This evening after leaving it on 6m ssb i came back into the shack to hear the tone and saw the disconnect box up on the screen. Had to long press the button to reboot the radio. I have now left it running on 6m wsjt-x monitoring fsk441 ms mode overnight to see how long it stays up this time.

I hope we are going to get a release to sort this issue soon - if thats not likely i think we need advice on which version to roll back to. Steve gw0gei / gw9j

Bill W2PKY

On April 8 I installed 1.10.16, the radio ran solid on Win7 Pro until today after applying Windows update KB4015552. The radio crashed after 2 1/2 hours. A long press was required to bring the radio back online and all 8 slices returned to their original configuration.
Coincidence??

UPDATE: Rolled back Win7 Pro "April roll up" and had to do an "OKAY Power on Reset" to clear the crashing problem.

Bill / VA3QB

I had my Maestro running today on 20M ....   I left the room and came back about 15 minutes later and it was at the screen where you select the radio but there was no radio to select.  I had to long press the start button to get a reboot then it was ok. This is version SmartSDR_v1.10.16   on a 6500.   The radio has DC removed every night. First loss of connection since install. I had been using  SmartSDR on my PC  a little bit earlier running JT-65.

73
Bill
VA3QB    

Jim Runge

My 6500 also locked up today with the tone and needing the long power button push. Latest software and only other thing running on the computer was the wooden box S meter. This is the second time in two weeks that the lockup has occurred.

James Skala

Has anyone seen the lockup occur while just using the Mastero and Flex

Dewey WØYWW

I've had 2 lockups running Maestro to 6700.  Each required reboot of 6700 to restart. Rebooting Maestro did not help. At the time of each lockup, rig had been sitting idle for less than an hour -- nothing that I can associate with probable cause.  Both Maestro and 6700 are connected directly to a 100mbs unmanaged ethernet switch (it has never ever lost a packet; always reads zero).  I only run CW and SSB, no digital.  Rig has been running ok for about 2 weeks since last lockup.  I'm using version 1.10.16.

Norm - W7CK

Watch the memory usage of SmartSDR.  See if it is slowly climbing out of control.

If this is what is causing the failure, I believe I read that the folks at Flex are aware of the memory leak and have fixed it.  The fix will be in the next software release.

If it is not related to the memory leak, then I guess it is an issue that still needs to be addressed.  

Bill / VA3QB

Yes

Steve Jones gw0gei

Surely the memory leak would eventually cause the pc to crash not the radio itself and would require the pc to be rebooted and not the radio? Either way, as an active contester on hf and vhf, a lack of reliabilty is a big issue. This is an issue that should be sorted in a maintenance release as soon as the solution is found, rather than being delayed for a new features or 2.0 release.

Norm - W7CK

I agree with that one!

Don w2xb

Just Had a lock up on 40 meters cw. In middle of a qso


Don...w2xb

NM1W

The top fixes should be fixing the radio crash, dax corruption, ssdr memory leaks...
This is  getting very old...

Walt

Thats for sure - very very old.  Had to reset DAX again to make WSJTx transmit tones.

here's hoping . .  

NM1W

6700 locked up just now during a wsjtx xmission; It locked up in xmit mode and required a long button push.......   radio had only been on about 30 minutes...    

Upon restarting smart sdr, it wasnt very smart about the slices. While I had 8 pans, I only had 4 slices randomly distributed, all were in cw mode before I had 4 cw, 4 digu.

James Skala

It happened again just flex and smartsdr.  I had just cycled with power off and within a hour it occurred.

Tim - W4TME

I have a few questions.

1.) Are you running 1.10.16?
2.) Do you have any USB cables connected to the radio?
3.) To reboot the radio did it take a press and release of the power button or a press and hold (5-10 seconds) before the radio started powering down?

James Skala

1.) Are you running 1.10.16?
***yes
2.) Do you have any USB cables connected to the radio?
***yes SPE 1.3k
3.) To reboot the radio did it take a press and release of the power button or a press and hold (5-10 seconds) before the radio started powering down?
***Long Press