namespace: implicitly adds DeviceAllow= when RootImage= is set

[thirdparty/systemd.git] / NEWS

diff --git a/NEWS b/NEWS
index 537c4b6131ad682fcb76bedfa6b2ce9a74b5453e..242f55af29d4c5303bb78a52274195c62800d3f4 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -1,5 +1,32 @@
 systemd System and Service Manager
 
+CHANGES WITH 240 in spe:
+
+        * A new service type has been added: Type=exec. It's very similar to
+          Type=simple and ensures the service manager will wait for both fork()
+          and execve() of the main service binary to complete before proceeding
+          with follow-up units. This is primarily useful so that the manager
+          propagates any errors in the preparation phase of service execution
+          back to the job that requested the unit to be started. For example,
+          consider a service that has ExecStart= set to a file system binary
+          that doesn't exist. With Type=simple starting the unit would
+          typically succeed instantly, as only fork() has to complete
+          successfully and execve() is not waited for, and hence its failure is
+          seen "too late". With the new Type=exec service type starting the
+          unit will fail, as the execve() will be waited for and will fail,
+          which is then propagated back to the start job.
+
+          NOTE: with the next release 241 of systemd we intend to change the
+          systemd-run tool to default to Type=exec for transient services
+          started by it. This should be mostly safe, but in specific corner
+          cases might result in problems, as the systemd-run tool will then
+          block on NSS calls (such as user name lookups due to User=) done
+          between the fork() and execve(), which under specific circumstances
+          might cause problems. It is recommended to specify "-p Type=simple"
+          explicitly in the few cases where this applies. For regular,
+          non-transient services (i.e. those defined with unit files on disk)
+          we will continue to default to Type=simple.
+
 CHANGES WITH 239:
 
         * NETWORK INTERFACE DEVICE NAMING CHANGES: systemd-udevd's "net_id"
@@ -82,7 +109,28 @@ CHANGES WITH 239:
 
         * systemd-resolved.service and systemd-networkd.service now set
           DynamicUser=yes. The users systemd-resolve and systemd-network are
-          not created by systemd-sysusers.
+          not created by systemd-sysusers anymore.
+
+          NOTE: This has a chance of breaking nss-ldap and similar NSS modules
+          that embedd a network facing module into any process using getpwuid()
+          or related call: the dynamic allocation of the user ID for
+          systemd-resolved.service means the service manager has to check NSS
+          if the user name is already taken when forking off the service. Since
+          the user in the common case won't be defined in /etc/passwd the
+          lookup is likely to trigger nss-ldap which in turn might use NSS to
+          ask systemd-resolved for hostname lookups. This will hence result in
+          a deadlock: a user name lookup in order to start
+          systemd-resolved.service will result in a host name lookup for which
+          systemd-resolved.service needs to be started already. There are
+          multiple ways to work around this problem: pre-allocate the
+          "systemd-resolve" user on such systems, so that nss-ldap won't be
+          triggered; or use a different NSS package that doesn't do networking
+          in-process but provides a local asynchronous name cache; or configure
+          the NSS package to avoid lookups for UIDs in the range `pkg-config
+          systemd --variable=dynamicuidmin` … `pkg-config systemd
+          --variable=dynamicuidmax`, so that it does not consider itself
+          authoritative for the same UID range systemd allocates dynamic users
+          from.
 
         * The systemd-resolve tool has been renamed to resolvectl (it also
           remains available under the old name, for compatibility), and its