Skip to content

June 24, 2014

2

PCoIP Log Viewer update for View 6.0

Let me open this post with an apology – I have been pulled away from supporting the Log Viewer (and the Config Utility) for far too long and correspondingly this site has suffered as well. So for all the users of these tools – I apologize.  I plan to devote time to both sets of tools in the coming weeks.  Thank you for hanging in there and finding value in these tools despite the absence.

Ok, with that out of the way-

What’s New

View 6.0 is here and with it – you guessed it – changes to the log file format.  In a previous post (here) I detailed a significant change to the log parser with the intent of making future parser updates much easier.  I am happy to say that investment is paying off. Minor changes were required to support both View 5.3 and View 6.0.  Previously, weaving in the new data detection and extraction would have required significantly more effort and would have been much more fragile.  As it is, things are working almost exactly as designed (I did make a few framework tweaks mostly around package visibility but I should have made those in the first place, so this was more of a correction than new development).

That being said, looking through the new files I am seeing a lot of information that is being reported differently, old information that was previously gone but is now back, existing data that is no longer there,  and some data that is entirely new. Some of this new data might turn out to be of great value. I’ll get into some of those differences in a moment, but it’s these discoveries that lead me to believe I’ll have at least one more follow-up release for View 6.0 once I confirm the value of this new data.

Reaching the Plateau – or Not

One change you will notice is that Plateau no longer seems to be tracked:

View 5.3 log file:

Tx thread info: bw limit = 1276, plateau = 1152.0, avg tx = 5.3, avg rx = 4.6 (KBytes/s)

View 6.0 log file:

Tx thread info: bw limit = 150, avg tx = 27.0, avg rx = 7.5 (KBytes/s)

So, to maintain simplicity, I am just setting Plateau to ‘0’ in the graphs.  From what I can tell, the value didn’t move as has happened in previous releases, it’s just not in there any longer.  If I hear  about or find some way to derive it I’ll add it back in.  Plateau was a good indicator of how much bandwidth PCoIP “thought” it could use at any given time so I do think some value has been lost here.

Other Changes

Loss conditions seem to have a completely new logging format with some new data not found in previous logs:

06/20/2014, 09:33:20.028> LVL:3 RC:   0 MGMT_PCOIP_DATA :ubs-BW-loss:  ----->  lost=1
06/20/2014, 09:33:20.047> LVL:3 RC:   0 MGMT_PCOIP_DATA :ubs-BW-tcp1:  ----->  lost/xmit=0.20/85.47, rate=0.2391
06/20/2014, 09:33:20.047> LVL:3 RC:   0 MGMT_PCOIP_DATA :ubs-BW-tcp2:  ----->  active[kbit/s]=12338.8543, TCP fair bw[kbit/s]=3317.7040, MSS=1100.00, RTT(aged)=124.1, loss=0.0391
06/20/2014, 09:33:20.047> LVL:1 RC:   0 MGMT_PCOIP_DATA :ubs-BW-decr: Decrease (loss) loss=0.001 current[kbit/s]=5819.2823, active[kbit/s]=12338.8543 -> 5734.3661, adjust factor=1.46%, floor[kbit/s]=104.0000

Tracing to the endpoint is also something I came across. It might be fun to build a network map out of this if the data is valid and usable. (IPs changed to protect the guilty):

06/20/2014, 09:24:58.026> LVL:2 RC:   0       SOCK_UTIL :trace_route: ---->  trace complete.  30 hops detected
06/20/2014, 09:24:58.026> LVL:2 RC:   0       SOCK_UTIL :hop [ 1] -- rtt [1]:     2 ms,  [2]:    <1 ms,  [3]:    <1 ms -- ip:   1.2.3.4 -- name: view.somehost.com
06/20/2014, 09:24:58.026> LVL:2 RC:   0       SOCK_UTIL :hop [ 2] -- rtt [1]:     2 ms,  [2]:     1 ms,  [3]:     2 ms -- ip:   1.2.3.4 -- name:            1.2.3.4
06/20/2014, 09:24:58.026> LVL:2 RC:   0       SOCK_UTIL :hop [ 3] -- rtt [1]:     2 ms,  [2]:     2 ms,  [3]:     1 ms -- ip:   1.2.3.4 -- name:            1.2.3.4
06/20/2014, 09:24:58.026> LVL:2 RC:   0       SOCK_UTIL :hop [ 4] -- rtt [1]:    34 ms,  [2]:    37 ms,  [3]:    35 ms -- ip:   1.2.3.4 -- name:            1.2.3.4

And lastly some new image stats that look VERY interesting:

06/19/2014, 12:09:01.723> LVL:2 RC:   0        MGMT_IMG :Image stats [MBits/s]: Total=0.428667 Lossy=0.120958 Lossless=0.242432 Build=0.0652768 (Lossy 0.0652768) Cache=0.000947985
06/19/2014, 12:09:01.723> LVL:2 RC:   0        MGMT_IMG :Image stats [MPPS]: Lossy=0.195575 Txt=0.162369 Bkgnd=0 Motion=0 TileCache=0.00347088 - deprecated - values duplicated below
06/19/2014, 12:09:01.723> LVL:2 RC:   0        MGMT_IMG :Image stats [MPPS]: Changed=0     LS(txt+bkg)=0.162369     LY(chg)=0.195876     LY(bld)=0.297299     LY(LS_res)=0
06/19/2014, 12:09:01.723> LVL:2 RC:   0        MGMT_IMG :Image stats [MPPS]: Motion=0     zero_offset_tile_cache_hits=0.00347088     offset_tile_cache_hits=0     temporal_cache_hits=0
06/19/2014, 12:09:01.723> LVL:2 RC:   0        MGMT_IMG :Image stats [MPPS]: LY_mec_restart=0     LY_txt_restart=0

First, it looks as if some previous stats like build/delta/lossy/changed have returned after getting lost somewhere around the 5.x time frame. Before I throw those back in, I need to validate them a bit more.  But more importantly, I believe this to be the first real appearance of CACHE stats in the logs! That might be very interesting! I have put in some requests to Teradici to help make sure I can interpret these values correctly and once I have that confirmed I suspect I’ll need to add a few new graphs in.

Getting the Update

If you already have the Log Viewer installed, you should get the update on the next launch. For those who do not have it, you will need to have Java 1.6+ installed and you can find the JNLP file here: http://hirstius.com/webstart/pcoip/PCoIPLogViewer.jnlp

As always, I expect there to be bugs.  I also have not validated the parser against EVERY possible build of View 6.0 though I did try to cover what I believe is the beta and the GA release. If you see an error saying no valid parser can be found, please reach out in the comments and I’ll arrange some way for you to send me the file so I can figure out what went wrong. Same thing for other errors I just don’t have a large sampling of data files to test against right now but I wanted to get this initial update out quick in the hopes that it helps at least some people right off the bat.

In closing, I want to say thank you again for your patience and I hope you continue to find this tool useful.

Read more from PCoIP, Tools, VDI, View, VMware
2 Comments Post a comment
  1. Kris Boutilier
    Jul 11 2014

    Looks like I’ve got an unsupported version. It’s being reported as:

    MGMT_SYS :Software Build ID: soft_pcoip_rc_3_12_APEX_view_5_3 3.12.3.28856
    MGMT_SYS :Software Build date: Dec 26 2013 10:27:35

    However it’s working just fine with our clients running:

    MGMT_SYS :Software Build ID: soft_pcoip_rc_3_12_3 3.12.3.26912
    MGMT_SYS :Software Build date: Oct 25 2013 13:35:15

    We’d be happy to share some sample logs.

    Reply
  2. Larry Blanco
    Aug 15 2016

    Any idea if you will add Horizon compatibility?

    Larry

    Reply

Share your thoughts, post a comment.

(required)
(required)

Note: HTML is allowed. Your email address will never be published.

Subscribe to comments