diff mbox

[nfs-utils] nfs-server-generator: avoid using external services.

Message ID 8737k0215d.fsf@notabene.neil.brown.name (mailing list archive)
State New, archived
Headers show

Commit Message

NeilBrown Oct. 13, 2016, 6:50 a.m. UTC
nfs-server-generator is run very early when a lot of services
are not yet started, so it mustn't depend on them.
Currently it can try to use hostname lookup and syslog.

Hostname-lookup is not needed, as we don't use the host issue,
and sending message to stderr is sufficient for the generator.

Disabling syslog is easy - call a function that sets a static variable.

Disabling hostname lookup isn't quite so easy, so add a static variable
which can be set to disable it.

Signed-off-by: NeilBrown <neilb@suse.com>
---

hi,
 I contemplated passing another arg to export_read() and export_d_read()
 to resolve this, but it didn't seem worth it.

 Is this approach OK to people?

Thanks,
NeilBrown


 support/export/export.c        | 12 ++++++++++--
 support/include/exportfs.h     |  1 +
 systemd/nfs-server-generator.c |  4 ++++
 3 files changed, 15 insertions(+), 2 deletions(-)

Comments

Steve Dickson Nov. 1, 2016, 6:45 p.m. UTC | #1
Hello,

On 10/13/2016 02:50 AM, NeilBrown wrote:
> 
> nfs-server-generator is run very early when a lot of services
> are not yet started, so it mustn't depend on them.
> Currently it can try to use hostname lookup and syslog.
> 
> Hostname-lookup is not needed, as we don't use the host issue,
> and sending message to stderr is sufficient for the generator.
> 
> Disabling syslog is easy - call a function that sets a static variable.
> 
> Disabling hostname lookup isn't quite so easy, so add a static variable
> which can be set to disable it.
> 
> Signed-off-by: NeilBrown <neilb@suse.com>
> ---
> 
> hi,
>  I contemplated passing another arg to export_read() and export_d_read()
>  to resolve this, but it didn't seem worth it.
> 
>  Is this approach OK to people?
It's not the most eloquent interface I've 
ever seen... ;-) 

What are the failures this is fixing? When a
hostname lookup is done what happens? Failure
hangs stopping the server from coming up?

 
> 
> Thanks,
> NeilBrown
> 
> 
>  support/export/export.c        | 12 ++++++++++--
>  support/include/exportfs.h     |  1 +
>  systemd/nfs-server-generator.c |  4 ++++
>  3 files changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/support/export/export.c b/support/export/export.c
> index 0b8a858c2c74..c65dc8807a4e 100644
> --- a/support/export/export.c
> +++ b/support/export/export.c
> @@ -67,6 +67,14 @@ static void warn_duplicated_exports(nfs_export *exp, struct exportent *eep)
>  	}
>  }
If we do have to stay with this approach, 
I definitely think comment explaining
why its needed and what it does.

Finally, on an unrelated issue I'm seeing 
SELinux preventing nfs-server-generator from 
accessing nfs-server.service.d/order-with-mounts.conf
because of a labeling issue. I guess the label
should be nfsd_unit_file_t not systemd_unit_file_t

steved. 
>  
> +static int assume_canonical = 0;
> +
> +void
> +export_avoid_host_lookup(int i)
> +{
> +	assume_canonical = i;
> +}
> +
>  /**
>   * export_read - read entries from /etc/exports
>   * @fname: name of file to read from
> @@ -83,9 +91,9 @@ export_read(char *fname)
>  
>  	setexportent(fname, "r");
>  	while ((eep = getexportent(0,1)) != NULL) {
> -		exp = export_lookup(eep->e_hostname, eep->e_path, 0);
> +		exp = export_lookup(eep->e_hostname, eep->e_path, assume_canonical);
>  		if (!exp) {
> -			if (export_create(eep, 0))
> +			if (export_create(eep, assume_canonical))
>  				/* possible complaints already logged */
>  				volumes++;
>  		}
> diff --git a/support/include/exportfs.h b/support/include/exportfs.h
> index f033329b5fd2..ba45b2dee199 100644
> --- a/support/include/exportfs.h
> +++ b/support/include/exportfs.h
> @@ -134,6 +134,7 @@ struct addrinfo *		client_resolve(const struct sockaddr *sap);
>  int 				client_member(const char *client,
>  						const char *name);
>  
> +void				export_avoid_host_lookup(int i);
>  int				export_read(char *fname);
>  int				export_d_read(const char *dname);
>  void				export_reset(nfs_export *);
> diff --git a/systemd/nfs-server-generator.c b/systemd/nfs-server-generator.c
> index af8bb52bfbd7..7ae6dd1c4b20 100644
> --- a/systemd/nfs-server-generator.c
> +++ b/systemd/nfs-server-generator.c
> @@ -93,6 +93,10 @@ int main(int argc, char *argv[])
>  	FILE		*f, *fstab;
>  	struct mntent	*mnt;
>  
> +	/* Avoid using any external services */
> +	xlog_syslog(0);
> +	export_avoid_host_lookup(1);
> +
>  	if (argc != 4 || argv[1][0] != '/') {
>  		fprintf(stderr, "nfs-server-generator: create systemd dependencies for nfs-server\n");
>  		fprintf(stderr, "Usage: normal-dir early-dir late-dir\n");
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
NeilBrown Nov. 1, 2016, 7:55 p.m. UTC | #2
On Wed, Nov 02 2016, Steve Dickson wrote:

> Hello,
>
> On 10/13/2016 02:50 AM, NeilBrown wrote:
>> 
>> nfs-server-generator is run very early when a lot of services
>> are not yet started, so it mustn't depend on them.
>> Currently it can try to use hostname lookup and syslog.
>> 
>> Hostname-lookup is not needed, as we don't use the host issue,
>> and sending message to stderr is sufficient for the generator.
>> 
>> Disabling syslog is easy - call a function that sets a static variable.
>> 
>> Disabling hostname lookup isn't quite so easy, so add a static variable
>> which can be set to disable it.
>> 
>> Signed-off-by: NeilBrown <neilb@suse.com>
>> ---
>> 
>> hi,
>>  I contemplated passing another arg to export_read() and export_d_read()
>>  to resolve this, but it didn't seem worth it.
>> 
>>  Is this approach OK to people?
> It's not the most eloquent interface I've 
> ever seen... ;-) 
>
> What are the failures this is fixing? When a
> hostname lookup is done what happens? Failure
> hangs stopping the server from coming up?

 I never actually reproduced it myself, but I think that when hostname
 lookup failed (possibly after a timeout) it tried to log an error via
 syslog and writing to syslog blocked indefinitely.
 The systemd.generator manpage explicitly rules out trying to talk to
 syslog.


>
>  
>> 
>> Thanks,
>> NeilBrown
>> 
>> 
>>  support/export/export.c        | 12 ++++++++++--
>>  support/include/exportfs.h     |  1 +
>>  systemd/nfs-server-generator.c |  4 ++++
>>  3 files changed, 15 insertions(+), 2 deletions(-)
>> 
>> diff --git a/support/export/export.c b/support/export/export.c
>> index 0b8a858c2c74..c65dc8807a4e 100644
>> --- a/support/export/export.c
>> +++ b/support/export/export.c
>> @@ -67,6 +67,14 @@ static void warn_duplicated_exports(nfs_export *exp, struct exportent *eep)
>>  	}
>>  }
> If we do have to stay with this approach, 
> I definitely think comment explaining
> why its needed and what it does.

That's sensible.  I'll resend with some nice friendly comments.

>
> Finally, on an unrelated issue I'm seeing 
> SELinux preventing nfs-server-generator from 
> accessing nfs-server.service.d/order-with-mounts.conf
> because of a labeling issue. I guess the label
> should be nfsd_unit_file_t not systemd_unit_file_t

I wouldn't know about that :-(

Thanks,
NeilBrown
Steve Dickson Nov. 1, 2016, 10:03 p.m. UTC | #3
On 11/01/2016 03:55 PM, NeilBrown wrote:
> On Wed, Nov 02 2016, Steve Dickson wrote:
> 
>> Hello,
>>
>> On 10/13/2016 02:50 AM, NeilBrown wrote:
>>>
>>> nfs-server-generator is run very early when a lot of services
>>> are not yet started, so it mustn't depend on them.
>>> Currently it can try to use hostname lookup and syslog.
>>>
>>> Hostname-lookup is not needed, as we don't use the host issue,
>>> and sending message to stderr is sufficient for the generator.
>>>
>>> Disabling syslog is easy - call a function that sets a static variable.
>>>
>>> Disabling hostname lookup isn't quite so easy, so add a static variable
>>> which can be set to disable it.
>>>
>>> Signed-off-by: NeilBrown <neilb@suse.com>
>>> ---
>>>
>>> hi,
>>>  I contemplated passing another arg to export_read() and export_d_read()
>>>  to resolve this, but it didn't seem worth it.
>>>
>>>  Is this approach OK to people?
>> It's not the most eloquent interface I've 
>> ever seen... ;-) 
>>
>> What are the failures this is fixing? When a
>> hostname lookup is done what happens? Failure
>> hangs stopping the server from coming up?
> 
>  I never actually reproduced it myself, but I think that when hostname
>  lookup failed (possibly after a timeout) it tried to log an error via
>  syslog and writing to syslog blocked indefinitely.
>  The systemd.generator manpage explicitly rules out trying to talk to
>  syslog.
This makes sense... 

> 
> 
>>
>>  
>>>
>>> Thanks,
>>> NeilBrown
>>>
>>>
>>>  support/export/export.c        | 12 ++++++++++--
>>>  support/include/exportfs.h     |  1 +
>>>  systemd/nfs-server-generator.c |  4 ++++
>>>  3 files changed, 15 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/support/export/export.c b/support/export/export.c
>>> index 0b8a858c2c74..c65dc8807a4e 100644
>>> --- a/support/export/export.c
>>> +++ b/support/export/export.c
>>> @@ -67,6 +67,14 @@ static void warn_duplicated_exports(nfs_export *exp, struct exportent *eep)
>>>  	}
>>>  }
>> If we do have to stay with this approach, 
>> I definitely think comment explaining
>> why its needed and what it does.
> 
> That's sensible.  I'll resend with some nice friendly comments.
Thanks! 

> 
>>
>> Finally, on an unrelated issue I'm seeing 
>> SELinux preventing nfs-server-generator from 
>> accessing nfs-server.service.d/order-with-mounts.conf
>> because of a labeling issue. I guess the label
>> should be nfsd_unit_file_t not systemd_unit_file_t
> 
> I wouldn't know about that :-(
I'll figure this out... 

steved.

> 
> Thanks,
> NeilBrown
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/support/export/export.c b/support/export/export.c
index 0b8a858c2c74..c65dc8807a4e 100644
--- a/support/export/export.c
+++ b/support/export/export.c
@@ -67,6 +67,14 @@  static void warn_duplicated_exports(nfs_export *exp, struct exportent *eep)
 	}
 }
 
+static int assume_canonical = 0;
+
+void
+export_avoid_host_lookup(int i)
+{
+	assume_canonical = i;
+}
+
 /**
  * export_read - read entries from /etc/exports
  * @fname: name of file to read from
@@ -83,9 +91,9 @@  export_read(char *fname)
 
 	setexportent(fname, "r");
 	while ((eep = getexportent(0,1)) != NULL) {
-		exp = export_lookup(eep->e_hostname, eep->e_path, 0);
+		exp = export_lookup(eep->e_hostname, eep->e_path, assume_canonical);
 		if (!exp) {
-			if (export_create(eep, 0))
+			if (export_create(eep, assume_canonical))
 				/* possible complaints already logged */
 				volumes++;
 		}
diff --git a/support/include/exportfs.h b/support/include/exportfs.h
index f033329b5fd2..ba45b2dee199 100644
--- a/support/include/exportfs.h
+++ b/support/include/exportfs.h
@@ -134,6 +134,7 @@  struct addrinfo *		client_resolve(const struct sockaddr *sap);
 int 				client_member(const char *client,
 						const char *name);
 
+void				export_avoid_host_lookup(int i);
 int				export_read(char *fname);
 int				export_d_read(const char *dname);
 void				export_reset(nfs_export *);
diff --git a/systemd/nfs-server-generator.c b/systemd/nfs-server-generator.c
index af8bb52bfbd7..7ae6dd1c4b20 100644
--- a/systemd/nfs-server-generator.c
+++ b/systemd/nfs-server-generator.c
@@ -93,6 +93,10 @@  int main(int argc, char *argv[])
 	FILE		*f, *fstab;
 	struct mntent	*mnt;
 
+	/* Avoid using any external services */
+	xlog_syslog(0);
+	export_avoid_host_lookup(1);
+
 	if (argc != 4 || argv[1][0] != '/') {
 		fprintf(stderr, "nfs-server-generator: create systemd dependencies for nfs-server\n");
 		fprintf(stderr, "Usage: normal-dir early-dir late-dir\n");