From 61c518627db9812d45d44c9813698da02e096e08 Mon Sep 17 00:00:00 2001
From: Paul Holzinger <pholzing@redhat.com>
Date: Mon, 28 Mar 2022 20:55:29 +0200
Subject: upgrade tests: fix networking problems

With podman4 we support netavark, however old versions will still use
cni. Since netavark and cni can conflict we should not mix them.
Remove the network setup from the inital podman command and create the
directories manually to prevent such conflicts.

Also the update to 4.0 changes the network db structure. While it is
compatible from 3.X to 4.0 it will fail the other way around. In this
test it will happen because the cleanup process still uses the old
podman while the network connect/disconnect test already changed the db
format. Therefore the cleanup process cannot see any networks and will
not tear it down. The following start will fail because the ip address
is already assigned.

Fixes #13679

Signed-off-by: Paul Holzinger <pholzing@redhat.com>
---
 test/upgrade/test-upgrade.bats | 41 +++++++++++++++++++++++++++++++----------
 1 file changed, 31 insertions(+), 10 deletions(-)

(limited to 'test')

diff --git a/test/upgrade/test-upgrade.bats b/test/upgrade/test-upgrade.bats
index 58e4fe0a3..063f4d624 100644
--- a/test/upgrade/test-upgrade.bats
+++ b/test/upgrade/test-upgrade.bats
@@ -105,7 +105,7 @@ podman \$opts run -d --name myrunningcontainer --label mylabel=$LABEL_RUNNING \
 
 podman \$opts pod create --name mypod
 
-podman \$opts network create mynetwork
+podman \$opts network create --disable-dns mynetwork
 
 echo READY
 while :;do
@@ -113,7 +113,10 @@ while :;do
         echo STOPPING
         podman \$opts stop -t 0 myrunningcontainer || true
         podman \$opts rm -f     myrunningcontainer || true
-        podman \$opts network rm -f mynetwork
+        # sigh, network rm fails with exec: "ip": executable file not found in $PATH
+        # we cannot change the images afterwards so we remove it manually (#11403)
+        # hardcode /etc/cni/net.d dir for now
+        podman \$opts network rm -f mynetwork || rm -f /etc/cni/net.d/mynetwork.conflist
         exit 0
     fi
     sleep 0.5
@@ -124,17 +127,14 @@ EOF
     # Clean up vestiges of previous run
     $PODMAN rm -f podman_parent || true
 
-
-    local netname=testnet-$(random_string 10)
-    $PODMAN network create $netname
-
     # Not entirely a NOP! This is just so we get the /run/... mount points created on a CI VM
-    # --mac-address is needed to create /run/cni, --network is needed to create /run/containers for dnsname
-    $PODMAN run --rm --mac-address 78:28:a6:8d:24:8a --network $netname $OLD_PODMAN true
-    $PODMAN network rm -f $netname
+    # Also use --network host to prevent any netavark/cni conflicts
+    $PODMAN run --rm --network host $OLD_PODMAN true
 
     # Podman 4.0 might no longer use cni so /run/cni and /run/containers will no be created in this case
-    mkdir -p /run/cni /run/containers
+    # Create directories manually to fix this. Also running with netavark can
+    # cause connectivity issues since cni and netavark should never be mixed.
+    mkdir -p /run/netns /run/cni /run/containers /var/lib/cni /etc/cni/net.d
 
 
     #
@@ -242,6 +242,8 @@ failed    | exited     | 17
 # if we can connect on an existing running container
 @test "network - connect" {
     skip_if_version_older 2.2.0
+    touch $PODMAN_UPGRADE_WORKDIR/ran-network-connect-test
+
     run_podman network connect mynetwork myrunningcontainer
     run_podman network disconnect podman myrunningcontainer
     run curl --max-time 3 -s 127.0.0.1:$HOST_PORT/index.txt
@@ -250,7 +252,26 @@ failed    | exited     | 17
 
 @test "network - restart" {
     # restart the container and check if we can still use the port
+
+    # https://github.com/containers/podman/issues/13679
+    # The upgrade to podman4 changes the network db format.
+    # While it is compatible from 3.X to 4.0 it will fail the other way around.
+    # This can be the case when the cleanup process runs before the stop process
+    # can do the cleanup.
+
+    # Since there is no easy way to fix this and downgrading is not something
+    # we support, just fix this bug in the tests by manually calling
+    # network disconnect to teardown the netns.
+    if test -e $PODMAN_UPGRADE_WORKDIR/ran-network-connect-test; then
+        run_podman network disconnect mynetwork myrunningcontainer
+    fi
+
     run_podman stop -t0 myrunningcontainer
+
+    # now connect again, do this before starting the container
+    if test -e $PODMAN_UPGRADE_WORKDIR/ran-network-connect-test; then
+        run_podman network connect mynetwork myrunningcontainer
+    fi
     run_podman start myrunningcontainer
     run curl --max-time 3 -s 127.0.0.1:$HOST_PORT/index.txt
     is "$output" "$RANDOM_STRING_1" "curl on restarted container"
-- 
cgit v1.2.3-54-g00ecf