Monday, September 18, 2017

Following the issues I had last week, I ended up updating to macOS High Sierra, twice. The first time was for the then current beta build, then the Golden Master (GM) was released, and I updated to that. That works relatively well, the only major issue so far being with VMware that reboots the system without asking when you try to boot Linux. Hmmm. Using VMware Fusion 7.1.2, that's quite outdated and may explain the problem.

THe good news is that I had no trouble with the build tools. I was afraid there would be some issue with new warnings or something like that.

Still did a few more experiments with RGS, updating configuration files and rebooting the guest. According to the documentation, you need to add the following to /etc/X11/xorg.conf:

Section "Module"
    Load "rge"
EndSection

But that module does not seem to run. I enabled logging to /var/log/rg.log using the /opt/hpremote/rgsender/senderConfigApp application, and I see the following (message already shown last week):

09-13-17 16:04:51 INFO  - BIOS UUID        : Exception caught. Failed to initialize WSTL. ACPI MCFG Exception. PCI Configuration space functionality unavailable

If I try to run rgsender_gui, I get this:

[root@f25-turbo rgsender]# ./rgsender_gui
./rgsender_gui: error while loading shared libraries: libIce.so.34: cannot open shared object file: No such file or directory

What is strange about this message is that libICE is something that my Fedora knows about, libIce is not.

I can run the shell script rgsender_gui.sh, and then I get the dreaded Invalid MIT-MAGIC-COOKIE-1, which I believe is X11-speak for a bad X11 authorization configuration.

Can't run rgsender either:

RGS Sender must be started by the Remote Graphics Extension

Oh well.

On the client side, it's clearly unhappy too, confirming the stuff is not running correctly:

Incompatibility between HP RGS and streaming agent? I see error messages that I don't like much (hand copied because copy-paste seems to have stopped working over the remote session):

UNKONWN msg of type 5
BAD VERSION 0 (expected is 1)
BAD VERSION 103 (expected is 1)
read command from device FAILED -- read 3 expected 8

And on next run:

write failed - Bad file descriptor
write failed - Bad file descriptor

Hmmm

Recorder update

Still mulling on the update of the recorder branch of the streaming agent after the conversion to C++ code. The part that causes me a bit of trouble is error handling, since in my experiments with frame rates, that had proved useful in detecting what the API would let me do and what was not OK.