- Progress on new timping features (send=, size=,...) - Warren
- Update on graphing problems at HEPNRC
- New Beacon sites - Les
- PPDG Monitoring - Les
- ESCC Presentations - All
- Action Items
It now records all the round trip times, not just the min/max/avg. Also modified
so that even if both the IP and name are provided it looks up the IP address from the name,
and only uses the IP address in the list if the name lookup fails.
Warren has also added the send= option to send other than 10 pings, size= to send other
than 100 and 1000 byte packets, he has also added a log= option that allows
one to select whether to log all pings. The new version is running at 2 sites.
The version that records all the data has been running at several sites for some time
and does not appear to have caused problems with the HEPNRC graphing program.
U Wisconsin ran into problems with Linux Redhat 6.1 in that the processes are not
properly cleaned up. They got around this by using the Nikhef ping instead.
However, this may not work for all sites.
It may be necessary to add the ability to kill processes to timeping.
Other suggested future options are: add ULM & XML formats; add Poisson scheduling.
George has taken a SAS course and now understands the code much better. He has
fixed most of the sites that wouldn't graph. The
non-graphing problems were not all the same. One more site will be
fixed by tomorrow. This leave us with the ESnet site that is password
protected (hence we don't know about whethere we can get graphs
- warren has provided the password), one
site that seems to be pinging only other local sites and therefore is
suspect, and 2 sites on which we're still working. Some of the problems may
be caused by having other than 100 or 1000 byte pings as is done at
New beacon Sites
A new list has been proposaed, see http://www-iwpm.slac.stanford.edu/pinger/beacons-mar00.txt
The following are comments from the list:
# --------------- New Beacons Updates March 2000 -------------------
# The last set of Beacons was finalized about 6 months ago.
# Now (March 2000) the Beacons need to reflect that have over doubled the number of countries
# monitored (72) and increased the # of monitoring sites and need to augment the Beacon sites.
# There have also been some corrections needed for the existing Beacon sites
# As of the release of this Beacon list we have 30 monitoring sites,
# 75 monitored countries, over 2200 pairs and 74 Beacons.
# The list below shows the changes made for the March 2000 Beacon list compared
# to the August-September 1999 Beacon list.
We are now awaiting the new timeping since several of the new Beacon sites
(e.g. Vietnam, Chile, Uruguay, Egypt) may have limited bandwidth.
There are 9 sites in the Particle Physics Data Grid (PPDG) (see
http://www-iepm.slac.stanford.edu/monitoring/ppdg/). PingER is already installed
at Wisconsin, SLAC, HEPNRC/FNAL, ESnet/LBL, BNL, ANL. We expect SDSC soon, are
working with Caltech, and have contacted JLab. When they are all running and we
have the new timeping we want to increase the ping frequency and make more detailed
correlations with throughput.
HEPNRC has not decided who will be attending the ESCC in Pleasanton yet. Probably both
Warren and Les will attend. They hope to present new information comparing various monitoring
methods (Surveyor/RIPE/AMP/PingER), performance ESnet vs Edu vs Com, International
Internet performance etc.
Further progress on prviding remote access to SAS requires the new server based on
NT. This will be focussed on after the new timeping is in production.
We agreed that Warren would make the current timeping available via the network
and tell George where to find it. He did this immediately after the meeting.
George will pick it up, look at and install.
When Warren gets back from vacation (April 5, 2000), we will look at how the
new timeping worked out with HEPNRC (it is already in use at SLAC & ANL) and
when appropriate (soon after) announce it to the monitoring sites.
The new beacon site list will be amended to take advantage of the size= option
to reduce the impact on some of the new less well connected sites, and then
distributed to the monitor sites.
George will then further review the timeping code, and look at the
requested additions and make estimates for the effort required. We will then discuss the
cost benefits of the proposed modifications and prioritize the work to be done.
Back to Top
Revised: march 21, 2000
Comments to email@example.com