Since ff8e1c37a5eb86b9cc45fb2b68c8fd33b79de64c landed, iSCSI installs fail. A crash happens near the end of the install process. The logs show this (not including all the context as the ultimate error is clear):
File "/usr/lib/python3.11/site-packages/dasbus/client/handler.py", line 483, in _get_method_reply
File "/usr/lib/python3.11/site-packages/dasbus/client/handler.py", line 509, in _handle_method_error
raise exception from None
pyanaconda.modules.common.errors.general.AnacondaError: NM.SettingIPConfig.get_dns_search() takes exactly 2 arguments (1 given)
Per a random copy of the API docs I found at https://lazka.github.io/pgi-docs/NM-1.0/classes/SettingIPConfig.html - I can't seem to find an 'official' copy - that function requires a parameter: "idx (int) – index number of the DNS search domain to return", but we're calling it just as `get_dns_search()`.
Proposing as an F39 Final blocker: "The installer must be able to detect (if possible) and install to supported network-attached storage devices", with footnote "Supported network-attached storage types include iSCSI, Fibre Channel and Fibre Channel over Ethernet (FCoE)."
The underlying C function here is https://gitlab.freedesktop.org/NetworkManager/NetworkManager/-/blob/main/src/libnm-core-impl/nm-setting-ip-config.c#L4242 . It has behaved the same way since it was added in 2014:
and the similar functions it replaced behaved the same too (they both required an index to be specified).
There is a `get_num_dns_searches()`, so we should probably call that first to find out how many search domains there are, then retrieve them all. Right now we're expecting `get_dns_search()` to return a comma-separated list, but it...doesn't. It never has.
diff --git a/pyanaconda/modules/network/nm_client.py b/pyanaconda/modules/network/nm_client.py
index 455caa5a2e..4e4c351b40 100644
@@ -1601,9 +1601,9 @@ def _update_ip4_config_kickstart_network_data(connection, network_data):
network_data.ipv4_ignore_auto_dns = s_ip4_config.get_ignore_auto_dns()
- ip4_dns_search = s_ip4_config.get_dns_search()
- if ip4_dns_search:
- network_data.ipv4_dns_search = ip4_dns_search
+ domains = [s_ip4_config.get_dns_search(idx) for idx in range(0, s_ip4_config.get_num_dns_searches())]
+ if domains:
+ network_data.ipv4_dns_search = ",".join(domains)
def _update_ip6_config_kickstart_network_data(connection, network_data):
@@ -1634,9 +1634,9 @@ def _update_ip6_config_kickstart_network_data(connection, network_data):
network_data.ipv6_ignore_auto_dns = s_ip6_config.get_ignore_auto_dns()
- ip6_dns_search = s_ip6_config.get_dns_search()
- if ip6_dns_search:
- network_data.ipv6_dns_search = ip6_dns_search
+ domains = [s_ip6_config.get_dns_search(idx) for idx in range(0, s_ip6_config.get_num_dns_searches())]
+ if domains:
+ network_data.ipv6_dns_search = ",".join(domains)
def _update_vlan_kickstart_network_data(nm_client, connection, network_data):
unfortunately I can't test the patch very easily in openQA because the test uses static networking and usually sets it up in the anaconda GUI, I'd have to remember the params to set it up on the cmdline so it could grab the updates.img...
Oops, sorry. But it's as you say - there are no good docs for this.
With https://github.com/rhinstaller/anaconda/pull/4614 as well, all should be fixed.
...or rather, I believe so.
Awesome. Once we get a Rawhide compose with the fixes in, openQA will let us know.
*** Bug 2176226 has been marked as a duplicate of this bug. ***