Age | Commit message (Collapse) | Author |
|
Change-Id: Iaf94de537f02186aee00b870c78828c0feae9038
|
|
Change-Id: I344147b0b47df20224952d34c46efd3eb233ec2c
|
|
Change-Id: I1231b7ba472a7b23f55f18f840f62eb9fc8616db
|
|
5 VLANs, spread across 5 switches
Change-Id: I3ab7ed68eec8cedf7ca38f5b0c57cb2076ce28a5
|
|
Change-Id: Ib592545e125561af707c72d6e308e20f2b5c71fd
|
|
Change-Id: Ie1ec8d05a81a684ce63ab8e254a46639ef7731be
|
|
Change-Id: I23850834563028a1d96b8d680fb0198a3e79cff7
|
|
Should be looking at VLAN_30, not VLAN_BASE
Change-Id: Ib1ac7c5eb8690a02cf60aa171249435e196d5c53
|
|
Remove vecho and vvecho, add vlog and vvlog instead.
log, vlog and vvlog are now wrappers around _log which handles verbose
levels better, for both terminal and log file.
Add LOGVERBOSE as a possible over-ride variable for a test script,
defaulting to VERBOSE + 1.
Change-Id: Iabba9b87de1ad1f8f4e287ed6574741dc7dc6e8f
|
|
Change-Id: I19aae78458e7a3ab0d47c0083730c18885fb9c44
|
|
Make it more suitable for logging sanely - don't print everything as
we go. Print out an overall PASS/FAIL message that others can parse
readily, and summarise the failures if any.
Change-Id: Idf2168bc9f8861fbe19c6a958bf0a6beaf1b6b4e
|
|
Change-Id: I9788d3f316ba4000ef61c5e778b2618ed7763478
|
|
Change-Id: I1a8475cad03e5d449f4093355e0aa5523d0107fc
|
|
Change-Id: Ifb1c9bbcd03f04f3f760ce6025920c4ed0f64fcd
|
|
Change-Id: I83a9446678978464622672fa3bfe20a7db28949b
|
|
Change-Id: I0cf6808ad222557db8e4d9d29044db5618e8e34f
|
|
We don't care too much about the numbers now. Any non-zero number of
failures is a FAIL.
Change-Id: I25a9ecfc1e69512c5bde8744bb8e54b7f3759646
|
|
Change-Id: I275f3254a2067522ecbf9cf63d622d0b50fd3207
|
|
Change-Id: Idd1c64d6ded5b2b6b6cf25de30954b390f3be6b8
|
|
Change-Id: I32ae4df770f20c256a2058dc44db58ee9d381a57
|
|
Rather than ignore the VLANd output, grab it and (optionally) log it.
Change-Id: I0dae6f4148f95e85f867f6fd8cfacf1aad7f92ff
|
|
Change-Id: I3867841dc53974e9d43c2fe6f33ba924c476d2f0
|
|
* Eat the output from any run_admin_command calls that weren't being
used
* Don't print the "full details" line by default
Change-Id: Ic5da2d9fb052fedf2574e7d4fa68f881ce7c8837
|
|
Change-Id: I4cec934c248a6bffedb338391d1c3f00430821d5
|
|
Only write any runtime test output to terminal on VERBOSE > 0
Change-Id: Ib540a3ec9aec7526b5a704192f985afecc969060
|
|
Change-Id: Icdb98881b8a8e6a2bdedab1fa327e559eeb506b7
|
|
Change-Id: I6e6dd6dc8f6216b7335f7180df0a84295d5b1bca
|
|
Add the test area definitions.
Add the final call to check_test_steps.
Remove the "sort -u $LOGFILE", we don't need to see that detail any
more by default.
Change-Id: I2f1af95581aded65d8731c36f946abe9b88a6757
|
|
Change-Id: Idfd2f426626ef537b3030a0518eb962d94245fa8
|
|
Rather than just dumping a set of ping results for the user to track,
add extra functions in the test-common wrapper.
For each test area in each test, mark the test area between
log "CHECK $AREA START"
and
log "CHECK $AREA END"
These will mark the interesting test areas, i.e. after VLAN changes
etc. have settled. Also, just after the START tag include some
statements like
log "CHECK $AREA CHECK FOO:machine1:machine2:machine3 BAR:machine4:machine5"
to define the expected connectivity results. FOO and BAR are machine
groups that are expected to see each other *only*, i.e. they've been
isolated using VLANs. At the very end of the test, add a call to
check_test_steps
to use this information and give a summary of what has happened in
each of the test areas. $AREA is a free-form string, just don't use
START, END, CHECK or whitespace within the name.
I've updated complex-4switches-2vlans-1 to use this new code, and it's
found another bug for me!
Change-Id: Ib41c7f1cc2fa8096c5579ab56d4199386abe6843
|
|
And fix up the banner comments to match reality
Change-Id: Ibddd2941cdc4874019f8e31cb7945884e99e6de2
|
|
Move over to the same style as the other tests, using $machine_PORT_ID
to pick up the right definitions from test-common
Change-Id: Idfdaa8eb6aaf2389d2542b848525d8e770f9d7c0
|
|
Use arndale03 instead of imx5303 which is misbehaving.
Change-Id: I187929a2ffa34cef5ce1fd311fba05afc7726701
|
|
Set up 2 VLANs across 2 switches each, each with 2 machines. Check for
isolation using 4 machines (1 on each switch again) remaining on the
base VLAN.
Change-Id: Iab68b860ef3779e303cc20d67915edfaa8578e67
|
|
Change-Id: If3017d5148472baa30c635879a47e74ab69171cf
|
|
test-3 tests 2 machines on vlandswitch01
test-4 tests 2 machines on vlandswitch02
test-5 tests 1 machine on each of vlandswitch01 and vlandswitch05,
with another machine on each switch watching too to validate
isolation.
Change-Id: I214fbc3165ec0c572865211241961fbc7d8b51e5
|
|
Change-Id: Ic0f71f6959bdc3adbf784669b93d9867c1e78986
|
|
Change-Id: Ied6203daea3e4c30b004883fb37b2504b4d0d35e
|
|
Change-Id: Idc67e08ed4287374476a16c04829e783e42be37d
|
|
Change-Id: Ic4a840314854b57a6acd7da9cb0698c5510acb0c
|
|
Change-Id: I656dd2a6c8334ab8fa3722f7b58e5c7551482c80
|
|
If we're being extra-verbose, we'll have multiple log lines per second
so we need more help to sort things properly.
Change-Id: I651a83250673d76372f89a2a7b8567f656de81e3
|
|
Add an extra level of logging for VERBOSE=2
* VERBOSE == 0 means "just log whatever the test script tells us"
* VERBOSE == 1 means "also log more intermediate logic"
* VERBOSE == 2 means "also log all the vland admin commands and
returns, plus all commands sent to the test
machines"
VERBOSE == 2 is very wordy, but may help debug failing tests
Move all uses of $($ADMIN foo) to use $(run_admin_command foo) to make
this easier
Change-Id: I0306ba4b11a1caacb18898de355be6310ab487bc
|
|
Match the timestamp format from the check-networks script, so they
sort properly
Change-Id: Ibe55c3798a75eac838458b9a6355778f4f99b2be
|
|
Change-Id: Ia7d72121882b67ebf7f14a3102a7e6f47f8eb03c
|
|
Change-Id: I69f5e85cec597aa304012900f84181e02e6a9105
|
|
Change-Id: I0660b6c8b07a4b434664ff6f635825fc05b20a39
|
|
Instead of running periodically out of cron and getting uneven
runtimes and logging, run continuously when we're told to. Use a
pidfile to protect against running multiple times.
Change-Id: I330f79f7325c7f3a0e22f9841833585e7164b242
|
|
Tests should now over-ride settings needed before they source the
test-common file: HOSTS, SWITCHES and VERBOSE.
Previously, the check_networks script would run automatically and exit
if told to using a flag file disable_test. Flip the logic: now the
list of hosts is passed into the check_networks script via a config
file instead, and if that config file does not exist then the
check_networks script will exit.
These changes allow for finer control of test setup: don't bother
pinging all hosts if we're just testing a couple of machines on one
switch, for example.
Change-Id: I0f0ef8ee83c0fffb40db1237048cb528494e881c
|
|
Otherwise things don't sort completely
Change-Id: I60e8485699df64110abb65127aba4e4b3413b6d7
|