X-Git-Url: http://git.onelab.eu/?p=bootcd.git;a=blobdiff_plain;f=ping-node.md;h=23a7d7c13196dc58fc07f52fb96d7c8378dd0d5d;hp=3048b5e3f09c9606a7797b66e8ccc07996fce338;hb=HEAD;hpb=8e9387ccf6bb0040f8f89a1d8fe256d27daeac09

diff --git a/ping-node.md b/ping-node.md
index 3048b5e..23a7d7c 100644
--- a/ping-node.md
+++ b/ping-node.md
@@ -13,24 +13,23 @@ And namely, the set of systemd files that describe booting off the CD image
 
 * and then run it on the testmaster side like this
 
-    iterate-ping-node <nb_iterations>
+  `iterate-ping-node -o run01 <nb_iterations>`
 
 * This allows to run a given bootcd (the iso computed for one node) several times over, and to gather all logs from qemu
-* This is **only** restarting the kvm/qemu node several times, nothing is done to recompute the .iso itself (see below for that). So the game is to easily simulate how a change to `bootcd` would affect a node ISO without rebuilding the whole damn thing.
+* This is **only** restarting the kvm/qemu node several times, nothing is done to recompute the .iso itself (see below for that). So the game is to easily simulate how a change to `bootcd` would affect a node ISO without rebuilding the whole damn thing
+* When -o is provided, the directory argument is created and all log files are stored in there
 
-## Easily redo a .iso
+# Easily redo a .iso
 
-* select a running test in testmaster/; like e.g. one that has failed the `ping_node` step already
+## preparation
 
-* you will need 2 local terminals with that
-  * one in `git/bootcd`
-  * one in `git/tests` 
+* select a running test in testmaster/; like e.g. one that has failed the `ping_node` step already
 
-* do the usual routine on running `exp`, exposing variables in both terminals,
+* you will need one local terminal in `git/bootcd`
 
-* then run `make sync` from workdir `tests/system` (just for `iterate-ping-node`)
+* do the usual routine on running `exp`, exposing variables in this terminal
 
-* and run `make sync-unwrap` from workdir bootcd
+* and run `make sync-unwrap` from this workdir `bootcd`
 
 At that point there will be the following files and subdirs on the KVM host (in my case boxtops)
 
@@ -43,9 +42,15 @@ At that point there will be the following files and subdirs on the KVM host (in
 * a read-only copy of the overlay image in `overlay.ref/`
 * a writable version of this in `overlay/`
 
-The workflow from then on is you can change the layout/contents of the bootcd/ directory, and then repeatedly run
+## iteration
+
+The workflow from then on is you can
 
-* `make sync-rewrap` from the `bootcd/` workdir, and then
-* `iterate-ping-node` from the `tests/` workdir
+* change the layout/contents of the `bootcd/` directory on the KVM host
+  * either manually right in the KVM host, and/or with
+  * `make sync-push` if you want to rsync the contents of `initscripts/` and ` systemd/` workdirs onto KVM 
+* and then rewrap the ISO image and hammer on it, and for this you run
+  * `make sync-rewrap` from the `bootcd/` workdir, and then
+  * `iterate-ping-node` from the `tests/` workdir
 
 Once you're satisfied you can make a difference between bootcd/ and bootcd.ref/ to see how the changes need to be implemented in `build.sh` and/or `prep.sh`